Skip to main content

Full text of "Readings in industrial organization and public policy"

See other formats


Digitized by the Internet Archive 
in 2013 

Readings in 





Volume VIII 

Selection Committee for This Volume 

Richard B. Heflebower 

George W. Stocking 

The participation of the American Economic Association in 
the presentation of this series consists in the appointment of 
a committee to determine the subjects of the volumes and 
of special committees to select the articles for each volume 


Industrial Organization 
and Public Policy 

Selected by a Committee of 


Published for the Association by 




First Printing, September, 1958 



Library of Congress Catalogue Card No. 58-11852 



This volume has two purposes: to provide for nonspecialists a con- 
venient compilation of the more important articles in the field of industrial 
organization and public policy that have appeared since Readings- in the 
Social Control of Industry was published in 1942, and to provide a book 
of readings useful in teaching graduate courses in this field. 

The field of industrial organization and public policy has neither a 
well-defined content nor precise boundaries. This became apparent to 
the editors when they began discussing the articles to be included in the 
volume. It became more evident as they received suggestions from econo- 
mists teaching graduate courses in this area. Lacking a coherent and 
unified content, this volume may not fit a particular graduate course as 
precisely as have other collections sponsored by the Association. 

The lack of unity in the articles selected reflects the variation in content 
of the graduate courses. Some courses emphasize the organization and 
characteristics of industrial markets in general and of some markets in 
particular. Others are designed to show the relationship between industrial 
markets and price theory and to modify and extend that theory in the 
light of the facts of industrial organization. Still others are concerned 
primarily with public policy issues, with little attention given to an 
analysis of market organization or to price theory. 

Reflecting somewhat this divergence of emphasis, the editors have 
classified the materials of this volume in five categories. They found it 
easier to establish the categories than to classify the articles. Nevertheless 
they believe that a grouping of articles by subject matter, imperfect 
though it may be, makes for a more orderly presentation of the materials 
than would a chance or arbitrary distribution. The categories selected 
are: The Structure of Industries and Markets; Case Studies in Industrial 
Structure and Behavior; Generalizations about Market Behavior; Indus- 
trial Organization and Economic Theory; and Competition, Monopoly, 
and Public Policy. 

This outline indicates how far the current conception of industrial or- 
ganization and public policy differs from that of Readings in the Social 
Control of Industry. The latter volume was concerned almost exclusively 
rX w i tn policy issues. Several of the articles dealt with the governmental 
problem of regulating prices, a topic omitted entirely from the present 
volume. Differences in the subject matter of the two volumes reflect 
differences in the content of graduate courses twenty years ago and today. 
Graduate courses are now concerned largely with the characteristics of 


markets of few sellers and the policy problems growing out of them. 
Policy problems are conceived as problems of antitrust rather than prob- 
lems of government regulation. But as antitrust problems they are con- 
cerned with an industry's organization and economic performance rather 
than with traditional legal concepts of antitrust administration. 

In determining the content of this volume the editors have received 
helpful suggestions from a number of specialists in the field. These were 
first asked to submit a list of articles which in their judgment should be 
included. The editors meanwhile had drawn up their own list. While 
numerous articles received the votes of two or more consultants, numer- 
ous other articles received only a single listing. This reflects both the 
diversity of emphasis in the content of graduate courses mentioned earlier 
and differences of judgment on the relative merits of articles. With the 
numerous suggestions before them the editors, relying on their own 
judgment where there was no agreement among consultants, drew up a 
tentative list for inclusion. This list, together with several specific titles 
which the editors were considering but on which they had not definitely 
decided, was submitted to consultants for comments and suggestions. The 
consultants' comments helped to narrow further the choice of articles. In 
the end the editors had to rely almost entirely on their own judgment in 
selecting articles representing a quarter or more of the pages reproduced 

In selecting or rejecting articles the editors established certain rules of 
eligibility. They decided against including sections of books, proceedings 
of conferences, and governmental hearings. A journal article that had been 
published in another book of readings was not automatically excluded. 
Articles representing a digest or survey of the literature, such as might 
appear in a professional law or business journal, were rated low by the 
editors. Preference was given to articles based on important research and 
on articles developing significant viewpoints on analytical or policy issues. 
Articles presented at professional meetings and subject to review and 
criticism by formal discussants provided a perplexing problem, for in 
some instances the discussant's comments were quite as significant as 
the original papers. Space limitations made it impossible to publish all of 
them. Recognizing that no one will be pleased with all our decisions, the 
editors have in some instances included such discussions but more fre- 
quently have not. 

Some who glance at the table of contents may be surprised that no 
article is included which makes use of "game theory" or "organization 
theory" in analyzing the conduct of firms in markets where sellers are 
few. No economist consulted by the editors incorporates these topics in 
graduate courses in industrial organization and public policy. This prac- 
tice may change. But with only limited space available, the editors de- 
cided to include only articles which reflect the more clearly neo-classical 


While the editors are grateful for the help of the consultants, they 
accept responsibility for the final selections made. They also gratefully 
acknowledge the painstaking work of Mrs. Elizabeth R. Post, a member 
of the economics department at Vanderbilt, and Hans G. Mueller, gradu- 
ate assistant, in compiling the bibliography. 

Richard B. Heflebower 
George W. Stocking 



1. The Measurement of Industrial Concentration .... 3 

By M. A. Adelman 

From The Review of Economics and Statistics, 1951 

2. Economies of Scale, Concentration, and the Condition of 
Entry in Twenty Manufacturing Industries 46 

By Joe S. Bain 

From The American Economic Review, 1954 

3. Monopoly and Oligopoly by Merger ... . ... . 69 

By George J. Stigler 

From The American Economic Review, 1950 


4. An Alternative Approach to the Concept of Workable 
Competition 83 

By Jesse W. Markham 

From The American Economic Review, 1950 

5. The Decline of Monopoly in the Metal Container Industry 96 

By James W. McKie 

From The American Economic Review, 1955 

6. The Tobacco Case of 1946 .105 

By William H. Nicholls 

From The American Economic Review, 1949 

7. The Cellophane Case and the New Competition . . . . 118 

By George W. Stocking and Willard F. Mueller 
From The American Economic Review, 1955 

18. B, 


8. Basing Point Pricing and Public Policy 153 

By Carl Kaysen 

From The Quarterly Journal of Economics, 1949 



9. The Nature and Significance of Price Leadership . . . 176 

By Jesse W. Markham 

From The American Economic Review, 1951 

10. Price and Production Policies of Large-Scale Enterprise . 190 

By Edward S. Mason 

From The American Economic Review, 1939 


11. Uncertainty, Evolution, and Economic Theory .... 207 

By Armen A. Alchian 

From The Journal of Political Economy, 1950 

12. A Note on Pricing in Monopoly and Oligopoly . . . . 220 

By Joe S. Bain 

From The American Economic Review, 1949 

13. Product Heterogeneity and Public Policy 236 

By Edward H. Chamberlin 

From The American Economic Review, 1950 

14. Competition: Static Models and Dynamic Aspects . . . 244 

By J. M. Clark 

From The American Economic Review, 1955; with excerpts from com- 
ment by Fritz Machlup and comment by Gardner Ackley 

15. Price Discrimination and the Multiple-Product Firm . . 262 

By Eli W. Clemens 

From The Review of Economic Studies, 1950-51 

16. The Influence of Market Structure on Technological 

Progress 277 

By William Fellner 

From The Quarterly Journal of Economics, 1951 

17. Toward a Theory of Industrial Markets and Prices . . . 297 

By Richard B. Heflebower 

From The American Economic Review, 1954 

18. Antitrust and the Classic Model 316 

By Shorey Peterson 

From The American Economic Review, 1957 


19. Public Policy and Business Size ^337 

By Corwin D. Edwards 

From The Journal of Business of the University of Chicago, 1951 


20. Standards for Antitrust Policy 352 

By Alfred E. Kahn 

From Harvard Law Review, 1953 

21. The Current Status of the Monopoly Problem in the 
United States 376 

By Edward S. Mason 

From Harvard Law Review, 1949 

Bibliography 393 



The Measurement of Industrial 


It is a resounding phrase, "the concentration of economic power," more 
often spoken than explained. By welding into a single expression the con- 
cepts both of a certain size structure and of a certain relation among the 
various parts of the structure, it suggests that instantaneous and transpar- 
ent connection between cause and effect which is all too rare in the im- 
perfect science of economics. Yet it appears also to have a certain poetic 
ambiguity. For one possible meaning of "economic power" is simply size 
— a corporation owning, say, $100 million of assets has, by definition, 
twice as much "economic power" as one owning $50 million. A second 
meaning is power over competitors or customers, power to insulate the 
firm against competition — in short, monopoly. 

Now the equating — by definition — of size with monopoly has oratori- 
cal advantages but may have analytical disadvantages. Our purpose is to 
explore concentration in the structural sense only, without regard to its 
behavior consequences; this paper is a study in the anatomy rather than 
the physiology of American' industry. Over the past twenty-odd years, 
considerable information has accumulated in this field, but knowledge is 
still considered unsatisfactory. 1 

A study of size structure is fairly distinct from a study of oligopoly in 

* The Review of Economics and Statistics, Vol. XXXIII (1951), pp. 269-96. Re- 
printed by courtesy of the publisher and the author. 

This article is a revision, considerably enlarged, of sections of a report submitted 
by the author last year to the Business Advisory Council of the Department of Com- 
merce. (The Council is an unofficial body, with no government status or budget.) 
Before any intensive research began, the Council granted the author full advance 
permission to publish any or all of the results as he saw fit, with the understanding 
that the Council was not necessarily in agreement with all of it or any of it. 
For this complete research freedom, my hearty thanks are due to the Council, and 
particularly to Blackwell Smith. Acknowledgment is also due to Murray F. Foss, 
Seymour E. Harris, Richard B. Heflebower, A. D. H. Kaplan, John Lintner, Her- 
bert C. Morse, and George J. Stigler, for helpful comments; and to John A. Menge, 
for skillful and devoted assistance. But none of those mentioned bear any responsi- 
bility for any error. 

t Massachusetts Institute of Technology. 

1 Report of the Joint Committee on the Economic Report, February 1949, p. 91. 



various markets, which is outside our province. 2 For the phenomenon of 
few sellers is conceptually different from the phenomenon of "big busi- 
ness": a small market may be occupied by a small number of small pro- 
ducers. This is not only a matter of arithmetic; a glance at Europe reveals 
business concerns generally smaller than in America, but markets smaller 
still, so that fewness is no less prevalent. 3 Our object, however, is not to 
measure the connection of fewness with size, but to measure size as such — 
the place of the big business units in the economy. 

In parts I and II, some conceptual problems of measurement are dis- 
cussed. Part III presents estimates of concentration in several important 
dimensions for the post-World War II period. Parts IV-VI test the hy- 
pothesis that concentration has increased during certain periods. Use is 
made of both primary and secondary sources, so that the article is in part 
a critical summary of the work of others, and in part an additional con- 
tribution, designed to undergo in time the same critical process. 

Concentration among business firms, like concentration of income or 
wealth among individuals or households, is fundamentally a measure of 
inequality of distribution. The usual practice in measuring inequality is 
to arrange the units in order of increasing size, and to express inequality 
in terms of percentages: thus the highest x per cent of all the units hold 
y per cent of, say, the total income of all the units. The comparison is 
most often made in terms of the familiar Lorenz curve, where percent- 
ages in terms of the possessing units are the horizontal axis, and per- 
centages of the dimension possessed are the vertical. Obviously, a diago- 
nal line from (0,0) to (100,100) expresses perfect equality, since at every 
point on the line the percentages are equal. The Lorenz curve is not only 
a simple and graphic presentation of the data; the ratio of the area be- 
tween the diagonal and the curve to the whole area serves as a coefficient 
of inequality. 4 

The Lorenz concept of inequality is fundamental, yet this brief de- 
scription should indicate its serious limits for our special purpose. For ex- 
ample, if an industry consisted of 1,000 firms, each with 0.1 per cent of 
the assets, it would show no more perfect equality than one containing 
two firms, each with one half. Tools of measurement should be adapted 
to the purpose of the inquiry, and it is obvious that our interest in the 
phenomena of big business and industrial concentration is not confined 
to the primary focus of great inequality of distribution; it is concerned 

2 William Fellner, Competition Among the Few (New York, 1949), pp. 17-24. 

3 The point is vigorously made for local American markets by Jesse W. Markham, 
"The Effectiveness of the Antitrust Laws: Comment," American Economic Review, 
XL (1950), pp. 167-69. 

4 The most usable form of the Gini coefficient which the author has found is 
described in Horst Mendershausen, Changes in Income Distribution during the Great 
Depression (New York, 1946), Appendix C-2. 


also with small absolute numbers. What we are really trying to describe 
are the very small number of very large firms, and their place in the 

The 200-odd concerns which are the largest employers are roughly 
one two-hundredth of one per cent of all business firms. In the presence 
of so extreme a degree of inequality, percentage distribution figures have 
very little meaning. The number of decimals becomes a serious matter, 
and the extent of rounding may lead to very great discrepancies among 
versions based on the same statistics.^ These arithmetical difficulties re- 
flect a deeper problem. It is no simple matter to decide how many units 
there are in the population, i.e., what constitutes a business firm. 6 (By 
comparison, the definition of a household is fairly simple.) 7 Large 
changes in classification of firms can be made, and large differences 
generated in the number of possessing units, with only negligible dif- 
ferences in the total amount possessed. Where so many of the units are 
peripheral and exert no market influence, percentages become even more 
ambiguous, and inter- temp oral comparisons virtually impossible. In re- 
cessions, the total number of business firms shrinks greatly; contrariwise 
during revival and inflation. 8 Therefore, if the number and holdings of 
the largest units remained constant, the decline in the total number of 
units would cause the largest units to become a relatively much larger 
fraction of all units. Hence Lorenz-type measurements would show a 
spurious de-concentration during recession, and an equally spurious in- 
creased concentration during revivals. 

One expedient is to assume that the total number of firms is constant 
throughout a time period, and equal to the earlier or later total, which- 
ever is smaller. The effect is to disregard firms entering and leaving the 
periphery of the business population. The procedure has much to recom- 
mend it, and has been used in two recent studies. 9 But by assuming the 
base population to be constant, one assumes in effect that a constant 
small number of firms is a constant percentage. For example, if the busi- 
ness population is four million firms in the earlier year of comparison, 
and three million firms in the later year, and the comparison is made of 
the holdings of the largest (say) 0.1 per cent in the later year, this 
amounts to comparing the share of the largest 3,000 firms in both years. 

5 Edwin B. George, "How Big is Big Business," Dun's Review, March, 1939, 
presents an amusing example. 

6 See Betty C. Churchill, "Revised Estimates of the Business Population 1929- 
48," Survey of Current Business, June, 1949. 

7 Cf. Mendershausen, op. cit., pp. 3-4; and the periodic Surveys of Consumer 

8 Cf. Churchill, op. cit.; and the Survey of Current Business, May, 1949, p. 3. 

9 John Lintner and J. Keith Butters, "Effect of Mergers on Industrial Concen- 
tration," Review of Economics and Statistics, XXXII (1950), pp. 46-47; G. Warren 
Nutter, A Quantitative Study of the Extent of Enterprise Monopoly in the United 
States (unpublished Ph.D. thesis, University of Chicago, 1949), pp. 43-46. 


Thus comparisons of a partition value, with the base population held 
constant, come to the same thing as comparisons of a small absolute num- 
ber. A Lorenz or Gini measure would give something more than this: 
the distribution among the three million. But it would not be free of 
ambiguity. Suppose the Gini coefficient to decrease because of greater 
equality of distribution among the lowest 99.9 per cent of all business 
firms; but suppose also that within the topmost group the largest 1,000 
firms gained at the expense of the other 2,000. One would hesitate to 
describe this as de-concentration in any meaningful sense. There is no 
practical way of avoiding this ambiguity, as long as our real interest is 
in the tiny uppermost portion of the Lorenz curve, except the inelegant 
procedure of making separate calculations for that group. Even here, the 
grouping would be essentially arbitrary, for there is no way of demon- 
strating that the top 0. 1 per cent was any better or worse a partition than 
the top 0.05 per cent or the top 0.15 per cent. In a word, there is no 
objective reason for taking the largest 1,500 rather than the largest 4,500, 
or vice versa. These are wide margins. So far, then, there is a funda- 
mental unclarity about just how small is the number of very large firms 
that we wish to study, whether in some particular industry or in the 
whole economy. 

Industry Concentration 

Measures of concentration for the whole economy are less popular than 
measures for particular industries. The well-known " concentration ra tio" 
for manufacturing industries is obtained by dividing total sales (or other 
dimension of size) of the largest four sellers by total sales of the whole 
group. The number four was chosen because the Census may not dis- 
close data for any smaller number of firms. The ratio is used for two 
rather different purposes, however. One is to measure the share of the 
largest four sellers of a particular product in a particular market, i.e. the 
extent of oligopoly. 10 (And oligopoly, as we have seen, has no necessary 
relation to size.) The imprecision of the market concept — Marshall's 
"industry" — is well known and needs none of the fatuous over-elabora- 
tion from which Stigler 11 and Fellner 12 have rescued us. 

The second purpose, which is ours, is to measure the relative size of 
the largest units, regardless of the degree of oligopoly. The two- or 
three-digit "industries" of the Standard Industrial Classification, which 
we shall use, are even more imprecise than the product groupings of the 
TNEC Monograph 27; but even if we made the ridiculously extreme 
assumption that firms within the same industry had no relations among 
themselves, i.e., that the two-digit industry was a purely random group- 

10 Thorp, Crowder, et aL, The Structure of Industry (TNEC Monograph 27, 
Washington, 1941), Part V. 

11 George J. Stigler, The Theory of Price (New York, 1946), pp. 280-83. 

12 Fellner, op. cit. y pp. 50-54. 


ing, it would still have a statistical usefulness. Industries would be a set 
of samples to be tested for concentration, and the mean result would be 
statistically much more reliable than a single estimate based on the whole 
population. It would indicate whether a change in concentration was 
characteristic of industry in general, or whether it had affected only cer- 
tain segments; it would supply the measure of dispersion without which 
a mean is of very limited use. Actually, the usefulness of industry group- 
ings is vastly greater than this. For even the most extreme opponents of 
the industry concept would need to admit that industries were rational 
subgroups into which to divide the population for enumeration; other- 
wise, they would need to repudiate nearly every statistical series now 
in use. 

A Suggestion for Measuring Concentration 

Z' The industry concentration ratio, and all the others based on small 
j numbers, suffers from the arbitrary element in the choice of numbers, 
and also wastes all the available information about the structure of the 
group itself. There is certainly a difference in structure between two in- 
dustries, each with a concentration ratio of (say) 0.50, but with the 
largest firm in one industry having 15 per cent of total output, and the 
largest firm in the second industry having 40 per cent. 

Now if we knew that firms A,B,C, etc., had respectively 20, 15, 10, 
etc., of the industry total, we can call x the number of firms, arranged 
from highest to lowest, and y the cumulative total. Then, as x equals 
1, 2, 3, etc., y equals 20, 35, 45, etc. (An example is the column of 
cumulative percentages in Table 4.) The curve described is concave 
downward. It starts from the origin, when x and y are zero; it approaches 
100 per cent, and equals it at some finite and sometimes very small value 
of x. The slope is always positive but always decreasing. A very simple 
form would be y = 1 — e' cx , where e is the Naperian constant. 13 This 
equation meets all our conditions, except that y can only equal unity, 
or 100 per cent, as x becomes indefinitely large. This seems to be of very 
little importance, since y can be made to approach unity so closely that 
the difference is negligible. 

The principal virtue of fitting such a curve would be not the curve it- 
self but the possibility of summarizing it in a single number. Professor 
John Lintner has suggested the use of the second derivative, which 
would express the rate of falling off; perhaps an even more intriguing 
possibility is the use of the exponent c as the coefficient of concentra- 
tion. In either case, all the information is used, the derivative or coeffi- 
cient can be calculated regardless of the size of the topmost group, and 

13 If y = 1 — e~ cx , the first derivative is -f- = ce~ cx , and the second derivative is 


dx>~ Ce ' 


comparisons can be made even when the groups being compared are of 
unequal size. 

Preliminary trials indicate that the equation y = 1 — e~ cx is not suffi- 
ciently flexible; at Massachusetts Institute of Technology, Professor Rob- 
ert Solow and Mr. William C. Hollinger have experimented with loga- 
rithmic transformations. One can only hope that much work will be 
done in this area, for much is needed. 


Thus far, we have discussed measures of distribution, and avoided the 
question of what is being distributed. There are four principal dimensions 
of size: employees, sales, income generated, and assets. 

Principal Dimensions of Size 

Employees. If one wishes to analyze the social and political aspects of 
industrial concentration, the distribution of employees is probably the 
most relevant single measure. Quite aside from this, number of employees 
is the only way of measuring changes in size (not concentration) over 
some time period. Monetary measurements are deficient not only because 
of price changes, for which allowance could be made, but because the 
significance of a given amount of output or assets is very different accord- 
ing to the general productive powers of society. If output per capita has, 
say, doubled over some time period, a firm which produces twice as much 
at the end as at the beginning can scarcely be said to have doubled in size. 
In society, man is still the measure of all things. The ultimate scarce fac- 
tor, as we learned during World War II and will doubtless learn again, is 

Sales. Sales provide the most easily available measure of size, but the 
weakest, in that they disregard the extent of vertical integration. Thus, 
two firms may each make ten per cent of an industry's sales; but if one 
merely purchases all the components, adds "a lick and a promise," and 
re-sells, it is obviously much smaller than the other, which undertakes 
all or much of the whole productive process. (Vertical integration is not 
relevant to concentration in the sense of the number of sellers in a 

Income Generated. The best single measure of economic size is in- 
come originating in the firm or industry, a concept almost identical with 
that of the Census 'Value added by manufacture." 14 Unfortunately, it is 

14 Value added is the total spread or margin between purchases (of raw ma- 
terials, fuel, containers, and power) and sales. A. D. H. Kaplan has pointed out to 
me that the 1941 Census of Manufactures, for the first time, excludes excise taxes, 
when they can be identified, from value added. Net national income also excludes 
indirect taxes; net national product includes them; both exclude depreciation. See 
the Survey of Current Business, National Income Supplement, July, 1947, p. 10(D) 
and Table 4. 


rarely available by firms, although it could easily be calculated if corpo- 
rate annual reports showed either (1) total purchases from other firms or 
(2) total wages and other income elements as well as profits. It would be 
even better to have both items and be able to cross-check. Corporate 
practice has improved in this respect, and it is to be hoped that despite 
the opposition of the public relations experts, 15 such statistics will become 
available for a sufficient number of firms to permit some useful computa- 

Assets. Assets reflect the d epth of produc tive activity in the firm in a 
manner analogous to income originating. If size is considered as the pres- 
ent result of forces operating in the past, assets are a better measure of size 
than income originating. Indeed, any real discrepancy between assets and 
income (aside from a statistical discrepancy which was due to imperfect 
data) would be not a difference in estimates of the same variable, but the 
difference between assets — the past absorption of resources — and produc- 
tion — the current addition to resources. A very familiar example of this ap- 
proach is the comparison of corporate profit with corporate assets or net 
worth — the rate of return. The ratio of total income originating to total 
assets would be a kind of index of productivity. Unfortunately, the data 
are too rough to permit such a computation; in this article, assets and in- 
come originating are assumed to be proportional. 

But although assets are thus an important measurement of business size 
and concentration, their use presents certain difficulties. Assets are the re- 
sult of accumulation over time. If prices have changed to an appreciable 
degree, comparisons among firms or groups of firms are impaired by 
variations in the time pattern of accumulation. The problem of estimating 
the valuation of income and assets during a period of rising prices has 
been exhaustively discussed. 16 Little more can be said here, except to point 
out the obvious: the longer the time period over which the assets have 
been accumulated, the greater the uncertainty about their valuation on a 
common basis. This uncertainty is widened still further by the variations 
in accounting and valuation methods among corporations; again, the 
longer the time period, the greater the room for error. 

An Example of Asset Concentration 

Since assets are used so frequently as a dimension of size and concen- 
tration, it is doubly useful to explore certain problems which their use 
involves by examining a recent and widely quoted report by the Federal 

15 For a typical argument see Don Knowlton, "Annual Reports as Public Rela- 
tions Tools," Journal of Accountancy, November, 1947; this may be profitably com- 
pared with an article in the same issue by J. M. Galanis, "The Security Analyst on 
Financial Statements." 

16 Much the best discussion is by Alexander, Bronfenbrenner, and Fabricant, in 
Five Monographs on Business Income, published by the Study Group on Business 
Income of the American Institute of Accountants (New York, 1950) . 


Trade Commission. 17 Table 1, which is based on that report, shows the 
percentage of the various types of assets held by manufacturing corpora- 
tions whose total assets in the first quarter of 1947 were over $100 million. 
The report used the category "net capital assets" as the best measure of 
size and concentration and applied it to 26 industries, despite the fact 
that this measure accounts for only one third of all assets. Some explana- 
tion is obviously necessary for this omission of two thirds of the informa- 

The omission of cash and cash equivalent (government securities) is 
not explained, except by a quotation from Gardiner C. Means 18 which 

Share of Assets Held by Certain Large Corporations, 

First Quarter of 1947 

Per Cent 
Held by 


Amount Held 

by All Manu- 


facturing Enter- 

with Assets 



Type of Asset 

(f Billion) 

$100 Million 


Cash and government securities 




Accounts and notes receivable (net) 





28.6 < 



Miscellaneous current assets 




Total current assets 






Net capital assets 


41.4- — 


Total tangible assets (3+6) 



Other assets (including deferred 





Total assets (5+6+8) 



* Includes unincorporated enterprises. 

Source: Federal Trade Commission, Report on the Concentration of Produc- 
tive Facilities, 1947 (Washington, 1949), p. 16. 

Note: Detail does not always add to total because of rounding. 

explains that the "ideal" figure for comparison uoith national "wealth 
would be that of tangible assets, i.e., capital assets plus inventories. But 
since the comparison here is being made, not with national wealth but 
with the assets of other business concerns, the omission of cash is a 
serious mistake. Economic theory and business practice both emphasize 
the importance of liquidity to the business firm. Cash is not only an im- 
portant asset; it involves no valuation problems. At any given moment, 
a dollar is worth the same in anybody's bank. 

The omission of receivables is explained only by an extremely brief 
reference to their "highly variable" nature. 19 What this probably means 

"Federal Trade Commission, Report on the Concentration of Productive Fa- 
cilities, 1941 (Washington, 1949) . 

18 Ibid., pp. 6-7. 

19 Ibid., p. 6. 


is that short-term payables and receivables can be increased together by 
almost the same amount, leaving the net position (in some sense) of the 
firm little changed. But borrowing on short term to acquire the means of 
accumulating receivables is no different in principle from borrowing on 
long term in order to acquire durable assets; and a certain normal 
amount of receivables is therefore a normal part of assets. If it is likely 
that both receivables and payables are temporarily inflated and will soon 
decline, one would welcome an attempt at adjustment, but there is no 
reason simply to disregard them. In point of fact, there has been no such 
decline in receivables since the first quarter of 1947, and this was obvious 
in 1949, when the report appeared. 

The exclusion of inventories — part of the "ideal" figure — from the 
measure of concentration is the strangest part of the whole procedure. It 
is explained as follows: 

According to reports of the Senate Small Business Committee, numerous 
small firms experienced great difficulty during that year [1947] in obtaining 
adequate inventories of raw materials. If inventories were thus relatively 
shorter in small than in large firms — as there is every reason to believe was the 
case — the inclusion of inventories in the measure would have the effect of over- 
stating the degree of economic concentration. 20 

Setting aside the question of the reliability of the evidence, our con- 
cern is with the logic of the procedure. A glance at Table 1 will show 
that the over-$100 million group held 35.6 per cent of inventories. Ac- 
cording to the report, this was an abnormally high percentage. In a 
more representative year, that is, it would be lower. Suppose we were 
to disregard this qualification for the moment, and add inventories to net 
capital assets to obtain net tangible assets — the "idsaT^figiire — as a meas- 
ure of concentration. Then the percentage held by the largest corpora- 
tions would be not 46.1 per cent but 41.4. But in a normal year, ac- 
cording to the report, the largest corporations would have held a smaller 
proportion of inventories than 35.6 per cent and therefore a smaller 
proportion of total tangible assets than 41.1 per cent. The FTC ex- 
planation can therefore be paraphrased as follows: toJiave used the esti- 
mate of 41.1 per cent would have resulted in an overestimate of the 
degree of concentration; therefore, we are using an estimate of 46.1! 

The report states that "capital assets consist of land, buildings, and 
equipment, which are physical, tangible, items." 21 "Capital assets" are no 
such thing. They are the record of money outlays on land, buildings, 
and equipment, over a fairly long period of time. The actual current 
value of the physical assets varies according to the time pattern over 
which they have been accumulated; and matters are made still more un- 
certain by adjustment for depreciation according to simple formulas 
whose chief virtue is convenience, and which vary among firms. In a 

20 Ibid., p. 7. 

21 Ibid., p. 6. 


Large Firms' Share of Total Employment, by Industry, 1948 

All Firms 


Number* ment* 
(Thousands) {Thousands) 

Firms Employing 10,000 or 

Per Cent 

Employ- of Total 

mentf Employ- 

Numberf (Thousands) mentj 

All Industries 




Food and kindred products 
Textile and textile products 
Leather and leather products 
Lumber and lumber products 
Paper and allied products 
Printing and publishing 
Chemicals and allied products§ 
Petroleum and coal 
Rubber products 
Stone, clay, and glass products 
Metals and metal products 
Primary metals 
Fabricated metals 
Machinery, except electrical 
Electrical machinery 
Transportation equipment 
Professional instruments, etc. 
Other Manufacturing || 
Public utilities 

Communication and other public 

utilities except transportation 
Wholesale trade 
Retail trade 

General merchandise 
Food and liquors 
Finance, insurance, and real estate 
Service industries 
























































































































































♦March 31, 1948. 

f December 31, 1948. 

t Percentages were calculated before rounding employment figures. 

n.a. = Not available. 

§ Includes Eastman Kodak and Celanese Corp. 

If Includes employment other than refining and' foreign employment. 

|| Includes tobacco and the Armstrong Cork Co. (classified under "Textiles" by the Department of Commerce), 
in addition to other miscellaneous manufactures. 

Sources: All firms from Murray F. Foss and Betty C. Churchill, "The Size Distribution of the Postwar Business 
Population," Survey of Current Business, May, 1950. 

For firms with 10,000 or more employees, the following master lists were consulted: (a) Department of Commerce, 
1000 Largest Manufacturing Corporations by Industry Groups (1948); (b) Securities and Exchange Commission, 
Survey of American Listed Corporations (1948); (c) Federal Trade Commission, Report ... on the Concentration of 
Productive Facilities, 1947: (d) Poor's Annual Register of Executives (1948); and (e) An unpublished table available 
by courtesy of A. D. H. Kaplan, presenting the 100 largest industrial corporations in 1948, by asset size. 

The following procedures were used: 

1. Mining, construction, and manufacturing. All firms which appeared in cither (a), (c), or (e) were checked in 
Moody's for their 1948 employment figures. Firms for which no 1948 figures were given were checked in Poor's. 
Source (b) was utilized by comparing sales, assets, and similar figures listed in this source with the sales, assets, etc., 
of those firms already examined which had barely attained the 10,000 employee level. The firms thus selected on this 
basis from (b) were then checked as before in Moody's and Poor's. Source (d) was used only as a check to discover 


word: the percentage of net capital assets is the best measure of concen- 
tration if and only if one wishes to maximize the probable error of the 


Tables 2-4 present estimates of postwar concentration in three impor- 
tant dimensions: employees, assets, and research personnel. The small top- 
most group includes in each case all firms which surpassed a given size in 
the particular dimension. The absolute numbers of the groups are not too 
greatly different — 260 in Table 2 A and 273 in Table 3 A. In view of the 
interest, which has become traditional, in the top 200 firms, calculations 
for them are also presented, but in brief. 

Employment and Asset Measures 

Tables 2A and 2B need little comment. The percentage figures for 
particular industries are often of ambiguous meaning because the large 
firms have employees outside the industry of their principal activity, as 
well as employees abroad. The most important cases are petroleum, 
chemicals, rubber products, and metals. Moreover, the percentages of 
Tables 2 A and 2B are somewhat inflated because December 1948 employ- 
ment was higher than March 1948. The largest 200 employers account for 
one fifth (19.8 per cent) of all employees in private nonagricultural es- 
tablishments, and for one eighth (12.4 per cent) of the total civilian 
labor force (Table 2B). 

Tables 3 A and 3B are concerned only with corporations. It will be 
observed that of the nearly four million business firms in existence by 
the end of 1947, less than one tenth took the corporate form. This under- 
lines in dramatic fashion the warning given above about the peripheral 
nature of most business firms, and the unreliability of percentages of the 

whether or not any obvious cases had been overlooked which might have 10,000 or more employees as given by 
Moody's or Poor's. 

2. Trade, service (includes motion picture producers and distributors), finance, and public utilities except trans- 
portation listings were obtained from (b), (d), and (e). The process of listing firms and deriving their employment 
data was, with one exception, exactly analogous to the process employed in obtaining the listings in mining, construc- 
tion, and manufacturing. The exception is that in several categories, such as "public utilities except railroads," and 
"finance," Poor's listings were small enough to permit a complete cross-check on employment data between Moody's 
and Poor's. 

3. "Public utilities-transportation" was taken directly from an unpublished table by Murray F. Foss, of the 
Department of Commerce, listing total employment figures for all railroads and similar carriers within different size 
classes, on the basis of number of employees per company. To this was added employment in airlines and local 
transit systems and similar carriers, obtained by procedures similar to those outlined above. 

4. One problem still remained. There were some firms which met the criteria of assets, sales, etc., but for which 
no 1948 employment figures were available in Moody's or Poor's. In the list of 260 firms employing 10,000 or more 
employees, such data were not available for 26 firms, of whom 14 also belong among the 200 largest. There were, 
however, previous employment figures, from 1940 to 1947, available for 13 of the 26 firms (and 8 of the 14). Estimates 
were then made from these figures, on the basis of the trend of employment for the industry, through 1948. This still 
left 13 firms in the largest 260 (and 6 firms in the largest 200) for which additional estimates were needed. This 
was done by ranking, on the basis of assets, sales, etc., these "unknown" firms relative to the firms in the same 
industry class for which employment figures were available. From their relative positions in these rankings, employ- 
ment figures were then interpolated. Errors in estimates of this sort are inevitable, but only 14 of the first 200 were 
involved in any estimating procedure, and for only 6 of these were there no employment data at all. Assuming, there- 
fore, as much as a 50 per cent under- or over-estimation for the whole group we would not find the over-all employ- 
ment figures altered by more than Yi to % of 1 per cent. Similar results hold for the largest 260 employers. Most of 
the estimates were for the class "manufacturing-textile products," and here the error may run as high as 7 to 10 
per cent. 


total number of firms. Table 3B embodies certain adjustments of Table 
3A which were possible only for very broad industrial classes. The pur- 
pose of the adjustments is to correct for some known inaccuracies, 
which tend to understate the degree of concentration, and which are ex- 
plained in detail below. 

But although the figures for individual industries are not satisfactory, 
they suffice to show that concentration is not only great but is itself 
highly "concentrated." Within the over-$100 million group, public utility 
corporations account for nearly half the total assets, and a single firm — 
the Bell Telephone System — accounts for one tenth. Among manufac- 

Share of Employment of Largest Employers, December 1948 

Per Cent 

Per Cent 


Employed by 

Employed by 


Largest 260 

Largest 200 





Total labor force 





Civilian labor force* 





Civilians gainfully employed! 





Private employment! 





Nonagricultural employment! 





Employees in nonagricultural estab- 






Employees, nongovernment! || 




* Excludes military, 
f Excludes unemployed. 
J Excludes government employment. 

§ Includes nonagricultural self-employed and government employment. 
If Excludes nonagricultural self-employed but includes government employment. 
|| Excludes all self-employed, but includes non-profit organizations and employees rendering 
professional services, in addition to employees of business concerns. 
Source: Table 2A and Survey of Current Business. 

turing industries, four industries — steel, automobiles, chemicals, and oil 
— account for nearly two thirds of the assets of the over-$100 million 
group. No more serious error could be made than to suppose the giants 
of American industry spread out evenly over its length and breadth. 

An estimate of total wealth held by the largest 200 corporations is 
desirable in order to compare over-all concentration in this dimension 
with concentration in employment (the "top 200" are not the same in 
each case). Our best estimate is that the largest 273 corporations hold 
46.3 per cent of total corporate assets. In order to eliminate the lowest 
73, an estimate of their holdings must be made. Let us for the moment 
make the arbitrary assumption that each of these 73 has just $100 million 
in assets. In that case, the group would account for $7.3 billion, which 
is just 3 per cent of total corporate assets, and the holding of the top 
200 would be 43.3 per cent (46.3 less 3.0). If we suppose that the average 
holding is not $100 million but $200 million, the lowest 73 would ac- 



count for $14.6 billion, or 6 per cent of total corporate assets, and the 
holdings of the top 200 would be 40.3 per cent. 

Table 3 A reveals that, of the over-$100 million group, there are eleven 
industrial groups, comprising 43 corporations, which average less than 
$200 million. It seems likely that in order to eliminate not 43 but 73 

Assets of Largest Corporations by Industry, 1947* 

Number of 



Total Assets 




Total with 



with Assets 

Col. 4 as 




Per Cent 

Industry Groups 


$100 Million 


$100 Million 

of Col. 3 






All industrial groups (except finance) 






Total mining and quarrying 






Total manufacturing 






Food and kindred products 


















Cotton manufactures 




Textiles, other 






















Lumber and timber 










Paper and allied products 






Printing and publishing 












Petroleum and coal 






Stone, clay and glass 






Iron and steel 






Nonferrous metals 






Electric machinery 






Machinery except electric and 







Autos and equipment 






Transportation equipment 






Other manufacturing 






Manufacturing not allocable 




Total public utilities 






Public utilities — transportation 






Public utilities — communication 






Other public utilities 






Total trade 






Total service 










Total agriculture, etc. 






Nature of business not allocable 



* Nonfinancial corporations. 

f "Public utilities — communication" contains largely members of the Bell Telephone System which should be 
counted as one firm. The total at the head of column (2) then becomes 273. 

Sources: Treasury Department, Press Service No. S-2449, Statistics of Income for 1947, Part 2, and unpublished 
data from the Bureau of Internal Revenue. 



Share of Assets Held by Largest Corporations, End of 1947* 

Corporations with Assets 
Exceeding $100 Million 

Total Assets, 
All Corporations 

Col. 3 as 
Per Cent 

Total Assets 

Industrial Group 


($ Billion) 

{$ Billion) 

of col. 4 








Total manufacturing 





Public utilities: communications 





All other 











Total manufacturing 





Public utilities: communications 





All other 










* Nonfinancial corporations. 

Sources: Unadjusted data from Table 3 A. 

Adjusted data: Manufacturing, see Table 11 below. 

"All other," adjusted by the factors developed for manufacturing. 

Communications: Federal Communication Commission, Statistics of the Communications Industry, 1947 
(Washington, 1948). The three corporations are the Western Union Telegraph Co., General Telephone Co., and the 
American Telephone and Telegraph Co. (Bell System). The last-named includes the Western Electric Co., which 
manufactures telephone equipment, and the total assets of all operating Bell System subsidiaries, although some of 
them are only partly owned by A.T.&T. The most important is the Southern New England Telephone Co., only 26.7 
per cent of whose stock is so held. It was impossible to make any estimates for International Telephone & Telegraph 
Co., and it was excluded from the table; its U.S. assets seem below $100 million. 

firms we would need to go well over the $200 million mark, and raise 
the average for the whole group to about $200 million. 22 This would 
mean that the share of the top 200 was 40.3 per cent. 

So much for the corporate universe. How large a part is it of the 
whole economy? The best comparison would be between the tangible 
wealth of the corporate universe and the total national wealth. However, 
this is impossible. Estimates of national wealth are still in a relatively un- 
developed stage; before the work of Kuznets and the National Bureau of 
Economic Research, 23 they were hardly respectable. Moreover, although 
paper claims could be stricken from both sides of the comparison, the 
item of land presents serious analytic difficulties. 

A simpler method is to assume that assets are proportional to income 
originating. In 1947, 58.8 per cent of total private income originated in 
corporate enterprise, 24 and we therefore assume that this represented a 
like share of total income-yielding wealth. In choosing this figure, we 

22 Since there would be 30 corporations with assets over $200 million to be elimi- 
nated, their higher average assets would be expected to balance their slight numerical 

23 Simon Kuznets, National Products Since 1869 (New York, 1946), pp. 185 If.; 
National Bureau of Economic Research, Studies in Incoine and Wealth, Vol. XIII 
(New York, 1950). 

24 Sjirvey of Current Business, July, 1950, p. 14. 


place consumer and government assets outside the universe in which 
corporate assets belong. Considerations of space forbid a defense of these 
two exclusions. 

If, therefore, 58.8 per cent of national income-producing wealth is 
held by corporations, and the largest 200 of these possess 40.3 per cent 
of the corporate holdings, then the "top 200" hold 23.7 per cent of the 
total. In language less precise but more accurate, we can say that the 
"top 200" hold between a fifth and a fourth of income-producing na- 
tional wealth. 

According to Table 2B, the largest 200 employers account for 15.4 
per cent of the total nongovernment labor force. Thus concentration in 
the dimension of asset ownership is about 60 per cent higher than in the 
dimension of employment; that is, employment among the top 200 is 
about 60 per cent more capital-intensive than employment in the econ- 
omy as a whole. 

Research Personnel 

At the end of 1945, according to the National Research Council, there 
were 2,443 industrial research laboratories employing 133,515 persons, 
of whom 54,321 were professional scientists or engineers. The Council 
estimated, however, that total personnel were underestimated by about 
5,000 because of the failure of some large corporations to answer their 
questionnaire. Thus, under the assumption that unreported personnel 
were divided between professional and unprofessional in about the same 
proportions as reported, it is possible to say, in round numbers, that there 
were 138,000 employees, of whom 56,000 were professionals. 

Table 4 shows the professional personnel of the largest 75 corporation 
laboratories, which account for nearly half the total. Close to one third 
of all professional personnel are employed by a score of the largest firms. 
It is clear t hat most of the organized industrial research is carried on by 
a very tiny fraction of all business firms, or even of all corporations. 
Moreover, the picture would not be greatly changed if the noncorporate 
research laboratories, governmental and nongovernmental, were in- 
cluded, for they employed only 4,300 scientists and engineers, fewer 
than the two largest corporate laboratories employed. 

A word of caution is needed in interpreting Table 4. The year 1945 
was obviously abnormal; specifically, it is not believed that normal re- 
search employment in the aviation companies, marked with an asterisk 
in the table, is as high as indicated by the 1945 data. Unhappily, 1951 and 
later years may prove to be equally abnormal. 


The outbreak of the Korean War, and the transition of this country to 
a more or less permanent state of semi-mobilization, have given a re- 
newed and very practical interest to the question: What happened to the 



Number of Professional Scientists and Engineers Employed 
in Largest Industrial Research Laboratories, 1945 









Per Cent 

Bell Telephone System 





E. I. duPont de Nemours and Co. 





Radio Corp. of America 





General Electric Co. 





General Motors Corp. 





Bendix Aviation Corp.* 





American Cyanamid Co. 





United Aircraft Corp.* 





Shell Union Oil Corp. 





Dow Chemical Co. 





Standard Oil Co. (N.J.) 





Monsanto Chemical Co. 





B. F. Goodrich Co. 





Socony-Vacuum Oil Co., Inc. 





Union Carbide and Carbon Co. 





Chrysler Corp. 





Hercules Powder Co. 





International Harvester Co. 





United States Steel Corp. 





The Texas Co. 





The Sperry Corp.* 





Lockheed Aircraft Corp.* 





Standard Oil Co. (Ind.) 





Philco Corp. 





Eastman Kodak Co. 





Pittsburgh Plate Glass Co. 





Phillips Petroleum Co. 





Standard Oil Co. (Calif.) 





Interchemical Corp. 





The Procter and Gamble Co. 





Sylvania Electric Products, Inc. 





Gulf Oil Corp. 





General Aniline and Film Corp. 





Westinghouse Electric and Mfg. Co. 





Remington Rand, Inc. 





Goodyear Tire and Rubber Co. 





Consolidated Vultee Aircraft Corp.* 





The Sherwin-Williams Co. 





Armour and Co. 





International Tel. and Tel. Corp. 

(Federal Telecommunication Labs. 






Merck and Co., Inc. 





Curtiss-Wright Corp. 





Johnson and Johnson 





Swift and Co. 





National Lead Co. 





The Atlantic Refining Co. 





Ford Motor Co. 





Ethyl Corp. 





Thomas A. Edison, Inc. 








TABLE 4 (Continued) 









Per Cent 

Lilly & Co. 





General Foods Corp. 





Aluminum Co. of America 





Continental Can Co., Inc. 





Celanese Corp. of America 





Reeves-Ely Laboratories, Inc. (Claude 

Neon, Inc.) 





Owens-Corning Fiberglass Corp. 





E. R. Squibb & Sons 





Vick Chemical Co. 





Armstrong Cork Co. 





Electric Auto-Lite Co. 





Sun Oil Co. 





Sinclair Oil Corp. 





Stromberg-Carlson Co. 





Upjohn Co. 





Reynolds Metal Co. 





Union Oil Co. of Calif. 





Continental Oil Co. 





American Can Co. 





Corning Glass Works 





Wyeth, Inc. (American Home Products 






Western Union Telegraph Co. 





Parke, Davis & Co. 





Johns-Manville Corp. 





Sterling Drugs, Inc. 





United Shoe Machinery Corp. 





Total — all corporate laboratories 




* Aircraft production. These companies are believed to have carried an abnormally large number of research 
employees in the year covered by the table. 

Source: Compiled from National Research Council, Industrial Research Laboratories in the United States (1946). 
The 1951 directory appeared too late for tabulation and inclusion here. 

pattern of concentration during World War II? On December 7, 1950, 
the Attorney General of the United States submitted a report to the 
President and Congress on the subject of "dangers to a competitive enter- 
prise economy which are inherent in mobilization for defense." 25 Two 
themes are apparent: one, the need to prevent collusive understandings 
under cover of industry defense planning; the other, the need for special 
favors to smaller business concerns in order to avert the possibility of 
increased concentration. The Attorney General makes the not unreason- 
able assumption that what happened during the last period of mobiliza- 
tion will probably happen during this one. On the basis of information 
supplied by the Federal Trade Commission, the report starts with the 
statement that "during the last war the long standing tendency toward 
economic concentration was accelerated." 20 Since increased concentration 

25 Letter of Transmittal attached to the Report of the Attorney General of the 
United States Prepared Pursuant to Section 708(e) of the Defense Act of 1950. 

26 Ibid., p. 4. 


is regarded as undesirable, it is considered necessary to counteract it 
even if this means that the Government will have to pay higher prices 
for defense supplies 27 or accept non-combat items of less reliable qual- 
ity. 28 

The wisdom of these or any other policies is beyond the scope of this 
article. Our concern is simply with the factual hypothesis: economic 
concentration in manufacturing increased during the period of the last 
war. Is this consistent with the available evidence, which may be itemized 
as follows: 

1. For the years 1941 and 1942, J. L. McConnell has adjusted corpo- 
rate rates of return for officers' compensation included in profits. Profits 


Effect of Taxes on Corporate Profits, By Size 

Groups, 1942 

Adjusted Rate of Return 

(in per cent 

of net worth) 

Total Assets Class 

($ Thousand) 

Before Taxes 

After Taxes 

Under 50 



























100,000 and over 



Source: Joseph L. McConnell, "1942 Corporate Profits by 
Size of Firm, " Survey of Current Business, January, 1946. 

before taxes for those years show little relation to size, except for a slight 
droop at both extremes, and a slight tendency to a maximum in the $5 
million to $10 million group. However, when data are given separately 
by industry groups, even this regularity disappears. 29 Profits after taxes, 
however, show a definite inverse relation with size. As Table 5 shows, 

27 Ibid., p. 39. 

28 Ibid., p. 36. The more euphemistic and also ambiguous original is that "efforts 
are being made to encourage the use of commercial or off-the-shelf items instead 
of technical military specifications wherever the rigors of actual military specifica- 
tions do not otherwise require." 

29 Joseph L. McConnell, "Corporate Earnings by Size of Firm," Survey of Current 
Business, May, 1945; "1942 Corporate Profits by Size of Firm," ibid., January 1946. 
McConnell's methods are not beyond dispute, and my personal belief is that they 
tend to overstate the profits of the smaller size groups, although not sufficiently to 
change the relation evident in Table 5. But Sidney S. Alexander, who used an en- 
tirely independent method, reaches the same conclusion for profits before taxes in 
1937. See "The Effect of Size of Manufacturing Corporation on the Distribution 
of the Rate of Return," Review of Econoviics and Statistics, XXXI (1949), pp. 


profits after taxes as a percentage of net worth were almost twice as high 
for the lowest as for the highest size group. Thus, for the war year 1942, 
one would expect smaller corporations, ceteris paribus, to be growing 
faster from retained earnings than larger, and concentration to be on the 

2. The Office of Price Administration compiled profits and assets of 
2,500 corporations, arranged by size classes, for the years 1939-44, which 
are summarized in Table 6. They show a rough inverse relation between 

Growth in Assets and Net Worth of 2,500 Corporations, 1939-44 

Number of 


Index Numbers 

(1936-1939 = 


Net Worth 

Total Assets 



1939 Asset 























{$ Million) 













Under 1.0 





















































































































500 and over 


























* Construction, mining, trade, services, and transportation (excluding railroads). 

Source: Office of Temporary Controls (OPA), War Profits Study No. 16 (Washington, 1947), Tables 23, 27. 

prewar size and subsequent growth of assets and net worth. In every 
case, the largest size classes grew more slowly than did the whole group. 

3. A larger OPA sample, confined to manufacturing, was subsequently 
published by the FTC (Table 7). The same relation is apparent for most 
but not all industry groups. The "$10 million and over" group grew 
more than the whole industry in 4 cases, grew less in 15 cases, and data 
were unavailable in 3 cases. 

4. The Federal Reserve Board, in collaboration with the Robert Mor- 
ris Associates, made a survey of business finance for the years 1940-44. 
They also found, as Table 8 shows, an inverse relationship between size 
and both profitability and rate of increase in assets. 30 It must be borne in 

30 See the following articles in the Federal Reserve Bulletin: 1945, F. C. Dirks, 
"Wartime Earnings of Small Business" and "Wartime Financing of Manufacturing 
and Trade Concerns"; Doris P. Warner, "Financial Developments in Manufacturing 
and Trade in 1944"; 1946, Albert R. Koch and Eleanor J. Stockwell, "The Postwar 
Financial Position of Business." 



Assets in 1945 of 22 Manufacturing Industries as Per Cent of Assets in 1941, 

by Asset-Size Groups 

Asset-Size Group ($ Mill 


Total, All 





10 and over 




125 N 















Textiles, except cotton 














































Printing and publishing 












Petroleum and coal 






Stone, clay, and glass 






Iron and steel 






Nonferrous metals 






Electrical machinery 






Machinery except electrical 






Automobiles and equipment 






Transportation equipment 






Other manufacturing 






Number of companies 






* Two companies of this group included in $1-5 million group, 
t Not shown to avoid possible disclosure of individual company information. 

Source: Federal Trade Commission, Report on Wartime Profits and Costs for Manufacturing 
Corporations (Washington, 1947), pp. 3, 19. 

mind that the FRB-A/Iorris Associates sample is very imperfect, as the 
first article in the series clearly explains. However, a later unpublished 
check of the estimates against the Statistics of Income showed a closer 
agreement than was expected by the careful authors of the study. 31 

5. The growth of assets of the 200 largest manufacturing corporations 
has been compared with that of 800 other corporations of medium to 
large size. For the period 1939-46, the former group increased its assets 
by 41 per cent, the latter by 102 per cent. 32 However, the value of this 
comparison for our purposes is somewhat weakened by the fact that the 
800 "other corporations," although chosen to be a "reasonably" good 
sample by industry and size, may yet be biased in some unknown way; 
the method of selection is not explained in detail. 

6. Data on the increase of stockholders' investment after 1940 in 
twenty-three selected manufacturing industries, as given in Table 9, show 

31 An unpublished comparison was examined by courtesy of Dr. Susan S. Burr 
and her staff at the Board of Governors of the Federal Revenue System. 

32 K. Celeste Stokes, "Financial Trends of Large Manufacturing Corporations," 

Survey of Current Business, November, 1947. 



Percentage Increase of Assets of Manufacturing and Trade 

Groups, by Asset-Size Groups 



Increase of 



Asset Size, End of 1941 ($ Thousand) 



250- 1,000- 
1,000 5,000 


10,000 and 


A. Manufacturing — War Industries 
201 242 222 188 
123 106 103 94 
247 256 229 177 


B. Man 

ufacturing — Other Industries 



139 126 
105 106 
146 134 

C. Wholesale Trade 





122 131 
105 105 

128 138 

D. Retail Trade 

5,000 and over 




135 140 
106 107 
143 150 


* Product of the two preceding lines, each based on a different sample. 

Sources: F. C. Dirks, "Wartime Financing of Manufacturing and Trade Con- 
cerns," Federal Reserve Bulletin, April 1945, pp. 1-18; Doris P. Warner, "Financial 
Developments in Manufacturing and Trade in 1944," Federal Reserve Bulletin, 
December 1945, pp. 1-6. 

that in 19 out of 23 industries, the percentage increase in the stockhold- 
ers' equity was greater for the smaller companies than for the top four 
in the industry. For the year 1947 (nearest to the war's end), the median 
increase among the big four was 42 per cent; for the others, 54 per cent. 

It is scarcely necessary, in a professional journal, to point out that all 
of the foregoing six items are imperfect and subject to varying margins 
of error. But the data all point in the same direction; the whole is greater 
than the sum of the separate items, 33 and indicates a slight de-concentra- 
tion during the wartime period. The only way to escape this conclusion 
is to assume a substantial wartime merger movement or wartime bank- 
ruptcies involving a considerable fraction of corporate assets. But mergers 
and bankruptcies were very few and the assets involved were insignificant. 

This is clearly incompatible with the sweeping statement upon which 
the Attorney General's report is based. We turn therefore to the source 
of that statement. The report was required under section 708(e) of the 
Defense Production Act of 1950, which states that the appropriate re- 
search may be done by the Federal Trade Commission. The Commission 

33 Cf . Morris R. Cohen and Ernest Nagel, An Introduction to Logic and Scientific 
Method (New York, 1935), pp. 282-84, 335, 349. 




Percentage Growth of Stockholders' Equities, Largest and Other 

Manufacturing Corporations, 1940-49 

Stockholders' Investment,* as Per Cent of 






Dairy products 

4 largest companies 




IS other companies 




Bakery products (bread) 

, . 

4 largest companies 




17 other companies 




Biscuits and crackers 

4 largest companies 




2 other companies 





4 largest companies 




4 other companies 





4 largest companies 




5 other companies 




Wool carpets and rugs 


4 largest companies 




9 other companies 




Linoleum and felt base 

2 largest companies 




3 other companies 




Paper and allied products 

4 largest companies 




100 other companies 




Industrial chemicals 

4 largest companies 




24 other companies 





4 largest companies 




3 other companies 




Soap, cleaning and polishing preparations 

4 largest companies 




4 other companies 




Petroleum refining 

4 largest companies 




36 other companies 




Tires and inner tubes 

4 largest companies 




12 other companies 




Flat glass, glassware (pressed and blown) 

4 largest companies 




7 other companies 




Abrasives, asbestos, and misc. nonmetallic mineral 


4 largest companies 




12 other companies 




Blast furnaces, steel works. 

, rolling mills 

4 largest companies 




30 other companies 




Primary smelting and refining nonferrous metals 

4 largest companies 




15 other companies 





TABLE 9 (Continued) 

Stockholders' Investment 

,* as Per 

Cent of 1940 





Engines and turbines 

4 largest companies 




11 other companies 




Tin cans and other tinware 

4 largest companies 




2 other companies 




Office and store machines and devices 

4 largest companies 




11 other companies 




Electrical machinery, equipment and 


4 largest companies 




^6L-Qther companies 




*Motor vehicles 


V__4 largest companies 




14 other companies 



Motor vehicle equipment 

4 largest companies 




37 other companies 




* Adjusted to exclude accelerated amortization on wartime facilities, 
t Erratic decrease not explained. 

Source: Computed from Federal Trade Commission, Rates of Return for 529 Identical Com- 
panies in 25 Selected Manufacturing Industries, 1940, 1947-49 (Washington, 1950). 

relied upon a single source for its statement of greater concentration dur- 
ing the last war: a Senate document 34 which, unlike the data cited up to 
now, was not incidentally but explicitly a study of wartime changes in 
the pattern of concentration. A careful examination of this report (here- 
after called the SWPC report) is therefore of methodological and prac- 
tical importance. 

The SWPC report appeared in January 1946, when items (1), (2), and 
(4) above had been published or their contents known to interested econ- 
omists. There is no reference to these three items, nor any reason given 
for not accepting them. The FTC report to the Attorney General men- 
tioned none of the six. 

The conclusions of the SWPC report rest on two propositions, which 
may be examined in turn. 

1. The report estimates that the 250 largest manufacturing corpora- 
tions held 66.5 per cent of total usable manufacturing facilities in 1945, 
including government-owned plants which they had an option to pur- 
chase if they so wished. In 1939, the identical corporations held 65.4 per 
cent of total usable manufacturing facilities. As proof of increased con- 
centration, this statement rests on (a) a logical fallacy, and (b) an unac- 
ceptable estimate. 

34 Economic Concentration and World War II, Report of the Smaller War Plants 
Corporation, 79th Congress, 2d Sess., Senate Committee Print No. 6. The following 
remarks are partly adapted from the writer's "Effective Competition and the Anti- 
trust Laws," Harvard Law Review, September, 1948. 


a. "These 250 largest manufacturing corporations are, for the most 
part, the traditional giants of American industry. But they include also a 
few corporations which have risen to positions of dominance during the 
war" 35 The error in this kind of reasoning is easily seen by observing 
that all corporations existing today own 100 per cent of the assets. These 
identical corporations owned a much smaller share than 100 per cent in 
previous years — many did not even exist. The increase to 100 per cent is 
scarcely a proof of greater concentration. Had the 250 largest in 1939 in- 
creased their share of the assets by 1945, or had the 250 largest in 1945 
held a larger share than the 250 largest in 1939, that would be another 
matter. The report avoids these two comparisons, and makes one which 
has no bearing on changes in the degree of concentration. 

b. "The report offers the estimate that about $20,000,000,000 of the 
$26,000,000,000 wartime plant is usable for the production of peacetime 
products, either immediately or after only minor reconversion. . . . This 
is simply an estimate of the portion of new facilities which are readily 
adaptable from a physical point of view." 36 Many products were needed 
in wartime in amounts far in excess of peacetime demand even under 
high employment. The estimate of the report is not the most probable 
value of usable facilities but the maximum value. A striking example is 
the aircraft industry, no less than 87 per cent of whose war-installed fa- 
cilities were considered "usable"; for manufacturing facilities as a whole, 
the comparable figure was only 77 per cent. 37 

Accordingly, the corresponding estimate of the SWPC report that 
$11.5 billion, or two-thirds of the $17.2 billion of federally financed fa- 
cilities held by the largest 250, would be usable, 3S must be drastically 
scaled down. In fact, the Civilian Production Administration had already 
estimated that only about one third of federally financed plants would be 
"readily usable in peacetime." 39 The SWPC report did not mention this. 
The two estimates may be consistent, since they do not refer to the same 
thing. But only the CPA figure is economically relevant. 

Therefore, if we make the appropriate adjustment, the largest 250 
manufacturing corporations in 1945 had $5.75 billion less in assets which 
they could take up at the war's end than the SWPC report estimated. If 
they took up all of it, their share of the total would then be barely over 
60 per cent — an actual decline of about 5 per cent since 1939. This does 
not support the hypothesis of greater concentration. 

2. "Small business was shoved into the background. Small firms (those 
with less than 500 employees) accounted for 52 per cent of total manu- 

Ibid., p. 40. Italics supplied. 


30 Ibid., p. 39. Italics in original. 

37 Calculated from ibid., Table D-l, p. 343. 
**lbid,:, pp. 38, 40. 

■ !i) Civilian Production Administration, War-Created Manufacturing Plant Federally 
Financed, 1940-1944 (November 15, 1945), p. 3. 


facturing employment in 1939. In 1944 this figure had fallen to 32 per 
cent." 40 But in fact as the small firms prospered, a sizable number of them 
"graduated" out of the less-than-500 class; and a little reflection will 
show that the more prosperous the condition of small manufacturers, the 
smaller the number and share of those remaining and the "worse" the 
statistical picture. Again and again, for each industrial group, the SWPC 
report rings all the changes on this elementary fallacy. One might as well 
argue as follows: in 1935-36, a large percentage of all families received 
annual incomes of $750 or less; today hardly any do; it was impossible 
to keep body and soul together on so little; therefore, this unfortunate 
class has been wiped out by the ravages of poverty. What the data indi- 
cate is an increase in average size. But they are compatible either with 
increasing or decreasing concentration, and therefore prove nothing one 
way or the other. 

The authors of the SWPC report had the opportunity to make a real 
contribution in a field important both in peace and war. Unfortunately, 
their data are simply irrelevant. There is no use crying over spilled milk; 
but there is every reason to urge that public policy be made on the basis 
of fact rather than warmed-over fiction. 


Let us now consider the trend of concentration over the period 1909-47. 
The year 1931 was a noteworthy one in the statistical study of concen- 
tration. First, the Statistics of Income began to present balance-sheet data 
classified by industry and size; second, there appeared an article by Gar- 
diner C. Means which may be said to have inaugurated the systematic 
study of concentration, and to which every student of the subject must 
be indebted. 41 For the first time, the topmost group of corporations was 
identified and placed in context. These findings were embodied in the 
better known book by Berle and Means, 42 published the next year, and 
elaborated some years later in a study of the National Resources Com- 
mittee. 43 The estimates precipitated a long controversy, often acrimonious 
yet often fruitful, which has been summed up with a light but masterful 
touch by Edwin B. George. 44 

Our chief concern at this point is with the trend of concentration 
rather than the level. References to increasing concentration are ex- 

40 SWPC report, op. cit., p. 24. 

41 Gardiner C. Means, "The Large Corporation in American Economic Life," 
American Economic Review, XXI (1931), pp. 10-42. 

42 A. A. Berle and Gardiner C. Means, The Modem Corporation and Private 
Property (New York, 1932). 

43 National Resources Committee, The Structure of the American Economy, 
Vol. I (Washington, 1939), particularly Appendices 10 and 11. 

44 Edwin B. George, "How Big is Big Business," Dun's Review, March, 1939; 
"Is Big Business Getting Bigger," ibid., May, 1939; "How Did Big Business Get 
Big?" ibid., September, 1939. 


tremely frequent. But nobody, to our knowledge, has stated the hypoth- 
esis with so much clarity as Berle and Means: 45 

Within "the corporate system" there exists a centripetal attraction which 
draws wealth together into aggregations of constantly increasing size. . . . The 
trend is apparent, and no limit is as yet in sight. Were it possible to say that 
circumstances had established the concentration, but that there was no basis to 
form an opinion as to whether the process would continue, the whole problem 
might be simplified. But this is not the case. 

Just what does this rapid growth of the big companies promise for the 
future? Let us project the trend of the growth of . . . the twenty years 1909 
to 1929, then 70 per cent of all corporate activity would be carried on by two 
hundred corporations by 1950. If the more rapid rates of growth from 1924 to 
1929 were maintained for the next twenty years, 85 per cent of corporate 
wealth would be held by two hundred huge units. It would take only forty 
years at the 1909-1929 rate or only thirty years at the 1924-1929 rates for all 
corporate activity and practically all industrial activity to be absorbed by two 
hundred giant companies. . . . Whether the future will see any such complete 
absorption of economic activity into a few great enterprises it is not possible 
to predict. . . . The trend of the recent past indicates, however, that the great 
corporation, already of tremendous importance today, will become increasingly 
important in the future. 

The Berle-Means projection was based on their estimate that the as- 
sets of the largest 200 were increasing during 1909-29 at the rate of 6.83 
per cent annually, while all nonfinancial corporations were growing at 
the rate of 3.72 per cent per year. Thus the top 200 were increasing 
relatively by 2.03 per cent per year. During 1909-24, the relative in- 
crease had been 1.66 per cent per year; during 1924-29 it had been 3.14 
per cent. 46 

Increased concentration during the 1924-29 period may be regarded 
as proved, i.e., the estimated change in the share of the top 200 seems to 
be greater than the reasonable limits of the dispersion around it. The real 
difficulty has been with the fifteen-year period 1909-24, and it has been 
best stated by George: 47 

The growth of both contestants was in fact understated because of the ab- 
sorption of water. Which of them suffered from that fact the most? . . . Until 
that surrealist equation is solved there is no precise answer to the basic question 
of who grew faster, and how much. 

Disputes touching the amount of water in 1909 capitalization and 1924 gross 
assets tend to throw us away from a measurement of finite bodies and toward 
one that could well bear the title A Few Speculative Thoughts on the Effect of 
Assumed Leverages on Unknown Weights. . . . 

We have little here but material for broad impressions. This is Dr. Means' 
own view of the matter, and he does not regard this particular stretch of his 
findings as other than reasonably indicative of an unequal rate of growth. 

There the matter rests. For what they are worth, the 1909-24 figures 
do show increasing concentration; we are not dealing with the kind of 

45 Berle and Means, op. cit., pp. 40-41. Italics supplied. 

,(i These are all compound rates. 

47 Edwin B. George, "Is Big Business Getting Bigger?" op. cit., pp. 36 and 56. 


irrelevancies reviewed in Section IV above. But the data seem far too 
weak to support much in the way of inference. 

This is not the case, however, with Means' later work for the National 
Resources Committee, which was based on unpublished income-tax data. 
For the years 1929-33, as Table 10 shows, there was a very appreciable 


Assets of Large Corporations as Per Cent of Total Assets* of 
Respective Groups, 1929 and 1933 

Corporations 1929 1933 

200 Large nonfinancial corporations 47.9 54.8 

75 Large manufacturing corporations 36.2 40.2 
85 Large transportation and other 

public utility corporations 79.0 86.1 

25 Large "other" corporations 10.9 14.8 

* Excluding nongovernment securities. 

Source: National Resources Committee, The Structure of the American Economy, 
Vol. I (Washington, 1939), Appendix 11, Tables IV. V. 

increase in concentration, both with respect to the top 200 as a whole and 
also its three main subgroups. The objection may, of course, be made 
that 1929-33 is not a "representative" period, as — thank heaven — it could 
not be. But this point really comprises two very distinct ones: (1) a sta- 
tistical peculiarity of assets during periods of rapidly changing prices, 
and (2) an assumption or prediction about future economic fluctuations. 
These may be considered in turn. 

(1) The top 200 had and have a disproportionately large share of the 
capital assets. Thus, while they held 47.9 per cent of total assets (less 
nongovernment securities) in 1929, they held 58.0 per cent of capital 
assets. Now capital assets consist of the historical cost of durable plant 
and equipment, while inventories consist of the current cost of non- 
durable assets. Given a period sufficient to turn over the capital assets, 
the latter come to reflect price movements. But after a short period of 
rapidly falling (rising) prices, there is bound to be the appearance of an 
increase (decrease) in the relative holdings of corporations that possess 
more durable assets. Hence the share of the top 200 in total assets less 
taxable investments is overstated as of 1933. A rough calculation suggests 
that from one third to one half of the 1929-33 increase in concentration 
is a statistical mirage. 

(2) But the rest of the increase is well established. Not only are the 
underlying data of high quality, but it also is well known that larger 
corporations make larger profits (or smaller losses) during an economic 
holocaust like 1929-3 3. 48 There is no question, therefore, that concentra- 

48 W. L. Crum, Corporate Size and Earning Power (Cambridge, Mass., 1939) 
especially Appendix Table B. 


tion must have increased during 1929-33, and that it may be expected to 
increase during any deep depression. 

The Trend of Concentration 1931-1941 

An attempt to plot the trend of concentration for later years runs into 
certain difficulties. One is the treatment of public utilities. During both 
the 1909-24 and the 1924-29 periods, this branch of the economy was no- 
table both for rapid growth and rapid concentration. For example, of the 
150 corporations included among the top 200 in both 1919 and 1928, the 
railroad group showed an increase in assets of 24 per cent, the industrials 
of 58 per cent, and the utilities of 194 per cent. For the hectic years 
1924-28, the annual rates of growth were, respectively, 2.3 per cent, 6.0 
per cent, and 15.9 per cent. 49 By 1930, according to a somewhat sketchy 
estimate, the three largest utility systems controlled about half the indus- 
try, and another ten controlled three fifths of the remainder. 50 In 1932, 
according to a better estimate, the operating companies controlled by the 
eight largest holding company systems generated 73 per cent of all pri- 
vately produced electric power. 51 In 1933, the largest 40 public utility 
corporations controlled over 80 per cent of the industry's assets. 52 

Under the Public Utility Holding Company Act of 1935, public util- 
ity holding companies had divested themselves of $16 billion of assets up 
to June 30, 1950. 53 This sum is over 70 per cent of the total assets owned 
by all electric and gas utilities in 1947. 53a Moreover, of the 41 electric and 
gas utilities which were among the 200 largest corporations in 1935, an 
incomplete count shows that at least 15, which accounted for well over 
half the assets, have been dissolved. 54 Ought this massive de-concentra- 
tion to be included in any measurement of trend? This depends funda- 
mentally on the context into which one wants to put the study. It would 
be necessary to include the utilities if our subject were the relation of 
concentration to both economic and political forces. Since our subject 
excludes the latter, it seems more appropriate to exclude public utilities. 

Moreover, there is general acceptance of the dichotomy between the 
competitive area of the economy, and the legal ("natural") monopolies. 
Of our 200 largest companies (whether by employment, assets, or re- 
search effort), about 150 are in the competitive area, and of these all but 
a score are in manufacturing. Hence an estimate of the trend of concen- 

49 Berle and Means, op. cit., p. 34. 

50 Charles S. Tippetts and Shaw Livermore, Business Organization and Control 
(New York, 1932), p. 509. 

51 Securities and Exchange Commission, Tenth Animal Report (Washington, 1945), 
p. 85. 

52 National Resources Committee, op. cit., p. 291, Table V-D. 

53 Securities and Exchange Commission, Fifteenth Annual Report (Washington, 
1950), pp. 62-63. 

o:!n Cf. Table 3A. 

54 National Resources Committee, op. cit., pp. 275-76, and SEC, op. cit., pp. 96-119. 


tration in manufacturing goes nearly all the way in meeting our need for 
estimates of concentration in this area. Moreover, should it turn out that 
concentration in manufacturing has been either stationary or declining, 
it becomes apparent that concentration for the economy as a whole must 
have decreased inasmuch as it has done so in the area of legal monopoly. 

In 1931, according to Statistics of Income, there were 139 manufac- 
turing corporations with assets of over $50 million, and they accounted 
for 46.5 per cent of the assets of all manufacturing corporations. But this 
must overestimate the number of corporations in the "over-$50 million" 
class, and it must underestimate their assets. For in some cases two or 
more subsidiaries of a large company made out separate tax returns, 
which caused that enterprise to be counted twice. Furthermore, there 
must have been separately incorporated subsidiaries which owned less 
than $50 million in assets, and these assets were therefore excluded from 
the total for the 139 largest. 55 

For 1947, as for no previous year, we are fortunate in having a reliable 
test of consolidation, and a means of correction. The Quarterly Indus- 
trial Financial Report Series, prepared jointly by the FTC and the SEC 
(hereafter called the FTC-SEC data), gives income and balance-sheet 
statements for all manufacturing corporations classified as to size, and 
these data are on a completely consolidated basis. (The data for the larg- 
est corporations rest not on sample estimates but on a complete count.) 
In Table 11 the FTC-SEC data for the end of 1947 are compared with 
those of the Bureau of Internal Revenue. At the end of 1947, according 
to FTC-SEC, manufacturing corporations with assets of over $100 mil- 
lion held a little more than 43 per cent of the assets of all manufacturing 
corporations. The agreement with the Treasury statistics is extremely 
close, and some congratulations are in order for the inter-departmental 
group of government statisticians who devised and launched the series. 
Line 7 of Table 11 shows the Treasury data as adjusted for consolida- 
tion. 56 Are there more or less in the top 1947 group than in the top 1931 

This question is not too difficult to answer. The Department of Com- 
merce has compiled a list of The 1000 Largest Manufacturing Corpora- 
tions as of the end of 1946. 57 At that time, there were, on a consolidated 

55 For the years 1935-41 inclusive, the privilege of filing consolidated returns was 
withdrawn from nearly all corporations except common carriers. Hence the cor- 
porate Statistics of Income for those years are not comparable with earlier or later 
years. But there was always some separate filing before 1935, and has been since 
1941, because the tax advantage may be overborne by other considerations. 

56 It would be preferable to adjust on the basis of total assets net of marketable 
securities, but the data are not available separately for the "over-$100 million" 
group. Our adjustment has the effect of removing marketable securities, calculating 
the adjusted percentages net of the type of asset, and then applying the percentages 
to all assets including marketable securities. This inflates the amounts but does not 
affect the ratios. 

57 Mimeographed, 1948. 



basis, 113 manufacturing firms with assets of $100 million or over. 58 Dur- 
ing the next year, some must have crossed the line, but how many? From 
the end of 1946 to the end of 1947, according to the Treasury, the assets 
of all manufacturing corporations increased by 11.2 per cent. 59 On the 
assumption that the corporations in the neighborhood of $100 million in- 
creased about as fast as manufacturing in general, it follows that a cor- 
poration with $86.5 million or more in assets in 1946 would be over the 


Estimated Asset Holdings of 139 Largest Manufacturing Corporations, End of 1947 
{All money figures in $ million) 

Total Assets 



Less Non- 














All manufacturing corporations* 

All manufacturing corporations with 

assets exceeding $100 million* 

Line (2) as per cent of line (1) 

All manufacturing corporations! 

All manufacturing corporations with 

assets exceeding $100 millionf 

Line (5) as per cent of line (4) 

Line (2), adjusted by line (6) (1.066) 

7 additional manufacturing corpora 


Line (7) plus line (8) 

Line (9), adjusted for wartime 

amortization % 

Line (1), adjusted for wartime 


Line (10) as per cent of line (11) 

Line (12), adjusted for abnormal 


27,634 33,351 60,985 100,654 111,356 

















) ... 













* Bureau of Internal Revenue. 

t Federal Trade Commission — Securities Exchange Commission. 

t See Table 12, Part A. 

§ See Table 12, Part B. 

Sources: Statistics of Income, 1947; Quarterly Industrial Financial Report Series, Fourth Quarter, 1947. 

$100 million mark in 1947. There were 19 such corporations on the Com- 
merce list. Hence it is probable that the over-$100 million class included 
132 corporations (113 plus 19) at the end of 1947. To reach a total of 
1 39, we need to add 7 more. In Table 1 1 the arbitrary assumption is 
made that each of the 7 are just barely under the $100-million mark, and 
should in the aggregate be credited with $0.7 billion of assets in all. This 
is of course an overestimate, but a very slight one. 

Two other adjustments are shown in detail in Tabic 12. The one for 
wartime accelerated amortization needs no comment. The other corrects 

58 Following the example of the FTC, we have added to the Commerce list two 
large firms whose assets were not made public. 

59 Statistics of Income for the respective years. 


for the abnormal type of asset structure discussed earlier (Section II). As 
Table 12 shows, the item of inventories was about 14 per cent of total 
assets in 1931, but nearly 25 per cent of total assets in 1947. It is obvious 
that because of the sharp price decline during 1929-31, inventories were 
abnormally low in the earlier year; and that because of the even sharper 
rise in prices through 1947, inventories were abnormally high in the later 
one. Since the largest corporations hold a smaller share of inventories 
than of total assets, this rise in the price level and in the value of inven- 


Explanation of Amortization and Inventory Adjustments 

A. Adjustment for Wartime Accelerated Amortization 

{All money figures in $ thousand) 


Total manufacturing, accelerated amortization, 




Total manufacturing investment thus amortized 

(line [1] X 5) 



Regular depreciation rate, 1944, in per cent 



Undepreciated per cent of (2) (100-[3 X line (3)]) 



Undepreciated portion of (2), end of 1947 

(line [4] X line [2]) 



Per cent of amortization in 1944 claimed by all cor- 

porations with assets exceeding $100 million 



Line (6) X line (5) 


B. Abnormal Inventories, 1947 

{All money figures in $ billion) 


Year 1929 1931 1941 



Ratio of inventories 

to total assets 17.95 14.3 23.5 



Factor, 14.3 -r- 24.8 



Total 1947 manufacturing inventories, adjusted 

($26.5 X 0.577) 



Total 1947 manufacturing assets, adjusted 

($97.6 less $11.3) 



Inventories of the largest corporations, adjusted 

($10.2 X 0.577) 



Assets of the largest corporations, adjusted 

($42.2 - $4.3) 



, Line (7) as per cent of line (5) 



, Adjustment factor, line (8) -4- 43.2* 


* For the source of this figure, see Table 11, line (6). 
Sources: Statistics of Income for the respective years. 

tories is a statistical distortion tending to understate concentration. Some 
upward adjustment is necessary in estimating concentration in 1947. Of 
course, were we to follow the suggestion of the FTC study discussed in 
Section II above, we would be compelled to make a downward adjust- 

As an approximation to a "normal" year, it is helpful to examine the 
ratio of inventories to other assets for other years of high peacetime em- 
ployment, as given by Statistics of Income. The results are inconclusive. 


For 1941, the ratio of inventories to total assets was 23.5 per cent, almost 
as high as 1947. But 1941 must also have been abnormal because of the 
defense boom and the shortage of materials apparent throughout. Hence 
the nearest year of high peacetime employment is 1929, when inven- 
tories were 17.95 per cent of total assets. However, to adjust both 1931 
and 1947 would involve greater uncertainties than to adjust either year so 
as to make it comparable with the other. Hence we have abandoned the 
attempt to estimate a normal ratio and have put 1947 on a 1931 basis; if 
inventories are adjusted accordingly (in effect, if the 1931 weights are 
substituted for the 1947), the share of the largest 139 corporations rises 
from 44.4 to 45.1 per cent. If we assume that the consolidation adjust- 
ment for 1931 (1.066) was the same as for 1947, then our best estimate 
is as follows: 

In 1931, the largest 139 manufacturing corporations held 49.6 per cent 
of the assets of all manufacturing corporations. In 1947, the largest 139 
held 45.0 per cent. (The unadjusted percentages would be 46.5 and 40.5, 
respectively.) This comparison covers sixteen years, but they were no 
ordinary years. As A4eans wrote at the halfway mark in 1939: 

An analysis of the economic structure ... is greatly aided by the depres- 
sion. The rapid drop in national production . . . and the very considerable re- 
covery since that time, give the economic analyst what is almost equivalent to a 
laboratory experiment on the basis of which structural characteristics may be 
observed . . . just as a high wind brings out the structural difference not evi- 
dent on a windless day between the tree that bends to the wind and that which 
stands unbending. 60 

These words are even more true today than in 1939, for since then we 
have had a strong wind blowing in the other direction: full and often 
overfull employment. As Table 13 shows, even seemingly slight tend- 
encies toward greater concentration would cumulate to formidable pro- 
portions within 16 years. Had the relative rate of growth estimated by 
Berle and Means for 1924-29 continued during this time, the unadjusted 
asset holdings of the largest 139 manufacturing corporations would in 
1947 have been over $85 billion, about three fourths of all manufacturing 
corporate assets. Even at the relatively low rate of growth for 1909-24, 
they would by 1947 have held well over 60 per cent of those assets. The 
rate of growth indicated over the entire period 1909-29 would have re- 
sulted in a figure between 60 and 70 per cent. Instead, the data show a 
reduction of nearly 10 per cent in the share held by the largest 139. 

That there seems actually to be a decline in concentration sounds more 
impressive than it should. Too many "bugs" remain in the statistics, and 
three further adjustments, at least, would be needed: (1) There were 
write-downs of assets during the 1930's, which were presumably more 
important among the large corporations. (2) There has been a long-term 

60 National Resources Committee, op. cit., p. 4. 


drift toward incorporation of an increasing fraction of business activity. 
But during World War II, a substantial number of smaller corporations 
became partnerships or proprietorships in order to avoid the excess prof- 
its tax; and not all had reverted to corporate status by 1947. Neither of 
these adjustments would be large; moreover, they would be in opposite 
directions, and the resultant would be smaller than either. (3) Possibly 
most important are the changes in industry classification since 1931, which 
might have involved large firms. However, our universe of manufactur- 
ing corporations is an extremely large one; and the reader has probably 
been struck by the minor percentage effects of large absolute adjust- 


Hypothetical and Actual Share of Assets of 139 Largest 

Manufacturing Corporations, 1931 and 1947 

Per Cent of 



Billions of 

All Mfg. 











Total assets, end of 1931 





Hypothetical total assets, end of 

1947, assuming rate of growth 

relative to all manufacturing 

corporations, in per cent per 


1 per cent 





2 per cent 





3 per cent 





4 per cent 





Actual total assets, end of 1947 


40. 5 f 


* Adjustments were made as follows: Line 1, for consolidation; lines 2 to 6, for consolidation, 
accelerated amortization, abnormal inventory holdings, and for the 7 additional corporations, 
t Includes 7 corporations added to the over-100 group as shown by Statistics of Income. 

ments. But let us waive any claims to precision, and summarize this long 
and arid inquiry: 

If there has been any strong and continuing tendency since 1931 to 
greater concentration in manufacturing, it must be detectable in the cor- 
porate balance sheet statistics for that period. These statistics do not 
show it. Therefore the tendency probably does not exist. 

If we suppose that the same economic forces were at work before the 
1930's as since then, these findings cast doubt on the Means estimates 
(outside of public utilities) for 1909-24. The two are not necessarily in- 
compatible, since they refer to different periods. But those who, unlike 
Means, have confidently extrapolated the 1909-29 tendencies forward, 
thereby implying that the two periods were substantially similar, are now 
in the embarrassing position of having to extrapolate figures back, and 



discovering that there has been no increase in concentration since 1909. 
But the safer procedure is simply to stick to the facts. 


Concentration in Manufacturing, 1901-41 

Toward the end of 1949, the student of concentration was blessed with 
two sets of data which could together be used to compare concentration 
over nearly half a century. One was a set of concentration ratios from 
the 1947 Census of Manufactures; the other was G. Warren Nutter's pa- 
tient archeological reconstruction of the period just after the turn of the 

Measurement of Concentration in Manufacturing Industries, 1901 and 1947 

Around 1901* 


Value Added 

{$ Million) 



with Concen- 

Per Cent 

Average (by 

tration Ratio 


Col. 1 as 

of Value 

Per Cent 

Value Added) 

over SO Per 


Per Cent 

, ofProd- 

of Value 

of Columns 

Major Industry Group 



of Col. 2 


Added % 

4 and 5§ 

(1947 Census Classification) 














Tobacco manufactures 






Textile mill products 







Apparel and related products 



Lumber and products (except 







Furniture and fixtures 



Paper and allied products 






Printing and publishing 






Chemical and allied products 






Petroleum and coal products 






Rubber products 






Leather and products 






Stone, clay, and glass 






Primary metal industries 





' 7.0 


Steel works and rolling mills 





Fabricated metal products 



Machinery except electrical 







Electrical machinery 




Transportation equipment 







Instruments and related products 



Miscellaneous manufactures 






Total all industries 







* 319 industries. 

t Per cent of total value of product of industry group accounted for by individual industries with concentration 
ratio over 50 per cent (440 industries). 

X Per cent of total value added by manufacture by industry group accounted for by individual industries with 
concentration ratio over 50 per cent (12 industries). .'.,-, , 

§ Column 6 substitutes a set of weights by value added for the system of weights by value-of-product which is 
given in Column 4. Hence Column 6 differs from Column 4 even when it is not necessary to average in the results of 

Sources: Cols. 1-3, G. Warren Nutter, A Quantitative Study of the Extent of Enterprise Monopoly in the United 
States, 1899 1959 (Unpublished Ph.D. thesis. University of Chicago, 1949). 

Cols 4~6, Letter from Charles Sawyer, Secretary of Commerce, to Emanuel Celler, Chairman of Sub-Com- 
mittee on Study of Monopoly Power, Washington, December 1, 1949, Table V appended; and Census of Manufac- 
tures, 1947. 


century. 61 Nutter's contribution is one of the most important in recent 

As Table 14 indicates, around 1901 nearly one third of the value added 
by manufacture was produced in industries where the concentration ra- 
tio was 50 or more. Since then, there has been both further concentra- 
tion and de-concentration. But the net result was that by 1947, only about 
one fourth of the value added was so produced. 

As usual, one must caution against hasty conclusions. In the first place, 
the skill of the investigator can only ameliorate his very faulty materials. 
Many of Nutter's sources are sketchy and unreliable, although they seem 
about as likely to be wrong in one direction as another. There seems no 
reason to impute any systematic bias to them; nevertheless, the standard 
error is inescapably large. 

Secondly, the concentration ratio itself has certain serious limitations, 
discussed in part earlier. The 50 per cent dividing line is, after all, an 
arbitrary one. Appreciable changes can occur in the extent of concen- 
tration without changing this ratio. For example, the share of the largest 
four might increase from just over 50 per cent to nearly 100 per cent, 
or from nearly zero to 49.9 per cent, without changing the ratio. How- 
ever, this is just where the industry-by-industry measurement of concen- 
tration is the most useful, in that it breaks up the whole class into many 
subgroups. It is highly unlikely that there could be many such shifts 
which carefully respected the 50 per cent line. With more than 350 ob- 
servations, one would expect such sampling fluctuations— for that is es- 
sentially what they are — to become quite small. Table 1 5 presents a f re- 


Frequency Distribution of Concentration Ratios over 50 Per Cent 
in All Industries 

Per Cent 

of Total 


Value Added 


Number of 


in All 




(f Billion) 


Per Cent 


























Total over 






Letter from 

Charles Sawyer, 

Secretary of Commerce, to Ema 

nuel Celler, Chair- 

man of Sub-Committee on Study of Monopoly Power, 


ington, Decern 

iber 1,1949, Table 

V appended; 

and Census 

of Manufactures, 

, 1947. 

61 Nutter, op. cit. Incidentally, his purpose, as the title indicates, was to measure 
the extent of monopoly rather than concentration, and his conclusion was that 
during 1899-1939 there was a slight increase in monopoly over the whole economy. 
Outside of manufacturing, there were no systematic data available, and he was 
compelled to separate sheep from goats in somewhat arbitrary fashion. Those who 
would reject such methods out of hand are urged to devise better ones. 


quency distribution of the concentration ratios in 1947. There is ob- 
viously no bunching at the highest levels, but an irregular decline from 
the 50-59.9 class upward. 

But other defects of Table 14 cannot be so clearly tested and dismissed. 
An increase in concentration via greater vertical integration might not 
be fully reflected in the concentration ratio. 62 The available evidence 
points to no increase in vertical integration by manufacturing establish- 
ments; in the absence of any other evidence, this would indicate no in- 
crease in concentration by firms; but the matter cannot be regarded as 


Increase in Number of Manufacturing Establishments, 

Number of 
Establishments Percentage 

Year (Thousands) Increase 

1899 204.8 

1904 213.4 

1914* 268.4 

1899-1914: 30 

1914f 173.6 

1939 173.8 

1947, adj.J 233.9 

1914-1947: 32 

1899-1947: 64 

* Old basis, i.e., counting all establishments selling at least $500 of 

t New basis, i.e., counting all establishments selling at least $5,000 of 

| According to the Census of Manufactures, 1947, Vol. 1, General Sum- 
mary, p. 4-5, sec. 8, "Change in Scope Between 1939 and 1947," it appears 
that the changes in classification brought about a gain of about 23,000 
establishments, and a loss of about 15,400, i.e., a net gain of about 7,600. 
This has been lowered to 7,000, to make allowance for the general growth 
of establishments since 1939, and subtracted from the 1947 total of estab- 
lishments to obtain an approximation to the total that would have been ob- 
tained on a 1939 basis. 

Sources: Census of Manufactures, respective years. 

settled. Finally, there can be significant differences in structure which 
any concentration ratio fails to reflect and which the exponent c, dis- 
cussed in Section I above, was designed to measure. Not even a rough 
adjustment can be made on this score; it is certainly a subject for later 

Both c and the concentration ratio measure concentration by the share 

62 Suppose the economy to consist of two industries, the first producing semi- 
finished goods which it sold to the second. One of the big four in the second might 
buy up (a) one of the big four in the first, resulting in a spurious decrease in the 
concentration ratio; (b) a firm not among the big four, resulting in a spurious in- 
crease in the concentration ratio. The mathematical theory, and the extent to which 
such distortions are overcome through value-added weighting, remain to be worked 
out. Were the concentration ratios themselves computed on the basis of value added, 
there would be no such problem, since value added reflects the degree of vertical 


of the small absolute number of the largest firms. But, the number of 
firms increased considerably during the half-century under study. A 
given number of firms became a progressively smaller per cent of all 
firms, and if the share of the smaller per cent in the later years was no 
less than that of the larger per cent in the earlier years then concentra- 
tion in some meaningful sense would have increased. In other words, for 
comparisons over such long time periods, we need some indicator of Lo- 
renz concentration, faulty as it is. Unfortunately, the basic figure for any 
such computation, the number of firms in manufacturing, does not exist. 
Let us, however, make shift with such data as we have on the number of 
manufacturing establishments, as summed up in Table 16, and assume 
that the number of manufacturing firms has also increased by about 64 
per cent. Now the 1901 data are divided into 319 industries and 1,276 
firms among the "big four"; the 1947 data have 452 industries and 1,708 
firms, an increase of 43 per cent, or much less than the increase in the 
total number of firms. Thus the number of the largest firms has de- 
clined, relatively speaking, to 87 (143 -=- 164) per cent of the 1901 level. 
But value added in industries where they hold at least half has declined 
more, to 76 per cent, 63 indicating a decline of Lorenz-type concentration 
of between 10 and 20 per cent. 64 

All the available evidence suggests that the number of firms has in- 
creased by less than 64 per cent, and therefore that the decline in Lorenz- 
type concentration is greater than just estimated. First, we have used the 
1899 figure for number of establishments, rather than trying to interpo- 
late for the 1901 figure, which would be larger. Second, if it is true that 
the average number of establishments per firm is larger today than at the 
turn of the century, as is generally believed, then the increase of manu- 
facturing firms must have been less than the increase of establishments. 
Third, the years after World War II contained a cyclical peak in the num- 
ber of firms. It is impossible that all of the new ones will survive. If we 
were able to make allowance for the normal shrinkage, the increase in 
the number of firms since 1939 (and 1899) would be much less. Our in- 
ability to make any precise estimate of the total number of firms is, as 
was indicated in Section I, the fundamental defect in Lorenz measure- 

Tables 14 to 16 appear to show a substantial decrease of concentration 

63 This amounts to the convention that the growth of the economy has been 
coterminous with the addition of new industries, which is of course not the case. 
However, the validity of the method is not affected by using this fiction. 

64 In an earlier draft, an error was committed which may also tempt some reader. 
The concentration ratio for the earlier years was multiplied by two factors: (1) the 
increase in the number of industries, and (2) the decline in average size of firm 
as the number of industries was increased. Such a procedure is valid in estimating 
the height of the curve controlled by an exponent of the type of c for any given 
year — it deals with a movement along the curve. The procedure cannot be used 
in order to decide how much the whole curve ought to be shifted to make some 
kind of allowance for the increased number of firms. 


in manufacturing since around 1901. But in view of the roughness of the 
early data and the crudity of some estimates, it seems best to state con- 
clusions as follows: The odds are better than even that there has actually 
been some decline in concentration. It is a good bet that there has at 
least been no actual increase; and the odds do seem high against any sub- 
stantial increase. 

The Trend of Concentration and the Role of Mergers 

The historian and the statistician use very different methods; funda- 
mentally they have the same task. Confronted with scattered pieces of 
evidence, each of them a more or less imperfect sample of the population 
concerning which information is desired, each attempts to draw up such 
hypotheses about the population — whether of events in time, or a distri- 
bution in space — as will not conflict with the data and may even enable 
one to predict what later pieces of evidence will look like. This article 
has been an attempt to explore a segment of American economic history 
since 1900 by way of the available statistical record, and that record is not 
plentiful. We have dipped into it at irregular intervals, and found no con- 
tinuing growth in concentration, but rather a surprising stability. It is 
clear that among nonmanufacturing industries, the railroads underwent 
both gigantic growth and great concentration before 1900; but they have 
not changed much with respect to concentration (and a few other things) 
since then. In the field of public utilities, Wall Street gave in the 1920's 
and SEC took away in the 1940's. As for the field of distribution, the rise 
of chain stores and mail-order houses since 1900 probably contributed to 
some increased concentration, although this can easily be exaggerated: 
most of the large distribution firms are department stores, and most of 
them are of a venerable age. Since 1929, as Table 17 shows, the chain 
stores and mail-order houses have not increased their share of the retail 
market. The field of financial and nonfinancial services is almost entirely 
unilluminated. This is not a very impressive body of evidence. 

Thus the field of manufacturing acquires a pivotal importance: be- 
cause it is large and a decisive trend in it would determine the trend for 
the economy; because it is the locus of such a disproportionate share of 
innovation and change; and because, not wholly without reason, we treat 
it as a sample of the whole economy. The following would seem to be a 
plausible sketch of developments there. In the nineteenth century, par- 
ticularly after the Civil War, there was a pronounced increase in the scale 
of manufacturing establishments and of firms, owing both to new tech- 
nical developments which made larger-scale production profitable, and 
perhaps even more to the rise of the railroad systems, the possibilities of 
nationwide distribution, and the hope of controlling the wider markets. 
Whatever the reasons, there occurred during the 1870's and 1880's not 
only growth of existing and new firms to larger size, but a great many 
mergers. The public reaction to "the trusts" was of course the Sherman 


Another wave of mergers came in the period of 1897-1903. Whether 
they were relatively more or less important than those of earlier years 
will perhaps never be known; but they were of extraordinary size. While 
the resulting increase in concentration cannot be measured, there is no 
question of the effect. 65 There is a compilation by the Census Bureau 66 of 
185 "combinations" formed up to June 30, 1900. The time period is thus 
very restricted: U.S. Steel, the greatest of all mergers, is excluded, and 
others as well. But the definition of a "combination" is also very narrow: 


Sales of Chain Stores and Mail-Order Houses Compared with 

Retail Sales, 1929-50 

Sales of 


Chain Stores 


and Mail- 

Col. 2 as 


order Houses 

Per Cent 

(f Million) 

{$ Million) 

of Col. 1 













































































Source: Survey of Current Business, various issues, 1943-51. 

two or more previously independent firms united by a charter obtained 
especially for that purpose, thus excluding purchases, exchanges of stock, 
and holding companies. Yet even this sadly truncated list accounted for 
8.4 per cent of all manufacturing employment in 1900. I doubt whether 
any such list, drawn up for any recent periods, would contain any oc- 
cupants. Nothing like this wave of mergers has ever been seen again. 

The last important wave of mergers was in 1924-29, hardly compara- 
ble to the earlier one, but still of substantial importance. 67 The Great De- 

65 J. Keith Butters, John Lintner, and William L. Cary, Effects of Taxation: Cor- 
porate Mergers (Boston, 1951), chap. x. 

66 Twelfth Census of the United States, 1900, Vol. VII, Part 1, pp. lxxv-xci. The 
number of combinations formed prior to 1897 was 65; to 1897, 7; to 1898, 20; to 
1899, 79; to 1900 (through June), 13. 

67 Means, op. cit., and Lintner and Butters, "The Effect of Alergers . . . ," op. cit. 


pression promoted further concentration, but the recovery and the war- 
time boom apparently worked in the opposite direction. 

After World War II there was another merger movement. The average 
annual number of mergers in 1946-47, as compiled by the Federal Trade 
Commission, was about one third as high as in 1929, and was exceeded 
by 10 out of the 13 years, 1919-31. Moreover, there were more firms in 
existence during the 1940's so that the rate of merger would need to be 
adjusted even further down. But of course the small number of mergers 
means very little. In the absence of data concerning the amount of assets 
absorbed, and by whom, it would not be proper to cite the low rate of 
mergers as proof that merger activity was not affecting concentration. 
The Commission, however, for reasons not explained in its report, in- 
terpreted the low rate of mergers as evidence that the mergers were 
actually "increasing concentration" and "strengthening the position of 
big business." 68 Its report concluded: 

... If nothing is done to check the growth in concentration either the giant 
corporations will ultimately take over the country, or the government will be 
impelled to step in and impose some form of direct regulation. . . . Crucial in 
that fight must be some effective means of preventing giant corporations from 
steadily increasing their power at the expense of small business. Therein lies 
the real significance of the proposed amendment to the Clayton Act, for with- 
out it the rise in economic concentration cannot be checked nor can the op- 
portunity for a resurgence of effective competition be preserved. 69 

The painstaking work of Lintner and Butters has shown that the 1940- 
47 merger movement had little or no effect on concentration. 70 The 
reader would do well to consult their article, and the later book, but the 
main conclusions can be summarized here. (1) Mergers of giants with 
giants, so prominent in the earlier merger movements, were unknown. 
(2) For all manufacturing and mining companies, the smaller the compa- 
nies, the more important were mergers as a source of growth. ( 3 ) Among 
the 1,000 largest manufacturing firms, the lower 500 (assets $7 million to 
$18 million) grew more proportionately by mergers than the upper half, 
i.e., there was some slight de-concentration through merger. (4) For the 
whole field of manufacturing and mining, the Gini coefficient (popula- 
tion assumed constant) increased through merger from .809 to .816, i.e., 
less than one per cent in eight years. 71 

So much for the negligible effects of mergers on the pattern of con- 

68 Federal Trade Commission, The Merger Movement: A Summary Report 
(Washington, 1948), p. 25. See also the Report of the Federal Trade Commission 
on the Present Trend of Corporate Mergers and Acquisitions (1947). 

eo jrp C, The Merger Movement: A Summary Report, p. 28. 

70 Lintner and Butters, "The Effect of Mergers . . . ," op. cit. 

71 Messrs. Blair and Houghton replied to Lintner and Butters, Review of Eco- 
nomics and Statistics, XXXIII (1951), pp. 63-67. They conceded all but the 
second point, and their attempt to defend even this one led to fresh errors. See the 
rejoinder by Lintner and Butters, ibid., pp. 67-71. 


centration. But whatever these effects, they were swamped, and sub- 
merged by other forms of growth. A generous estimate is that not over 
$5 billion was involved in all manufacturing and mining mergers during 
1940-47. 72 But during this period, according to a source which is biased 
downward, the total assets of all corporations in these fields increased 
from $67.8 billion to $118.6 billion; 73 the increase was over ten times the 
amount involved in mergers. As we saw earlier, no growth of concentra- 
tion is visible either in 1931-47, or in 1939-47. 

The statement of the Federal Trade Commission is in part a policy 
judgment, with whose wisdom we are not here concerned. 74 It is also a 
statement of fact, and as such it is doubly wrong: there has been no in- 
crease in concentration, and mergers have not been important enough to 
be of any effect, one way or the other. 


The most obvious conclusion is also the most depressing one: how lit- 
tle we know of our industrial structure and its evolution. At the begin- 
ning of this paper, we deliberately renounced all the more difficult prob- 
lems of economic behavior and aimed only at description. Yet even the 
simplest problem could not be approached without all manner of approxi- 
mations and expedients which quickly dissolved any hope of precision. 
Not only are the most important basic data not available; we have scarcely 
even begun to decide what questions we want answered. 

Need for Further Research 

Work is needed in three directions: (1) The. meaning of concentration 
and the development of indices of the degree of concentration. Earlier 
we indicated one type of curve which could be developed for this pur- 
pose, but the surface has scarcely been scratched. (2) The historical de- 
velopment of concentration and de-concentration in specific industries. 
This should not only be of considerable intrinsic interest but should con- 
tribute to our better understanding of the techniques of measurement 
and the economic determinants of concentration. (3) Measurement of 
the place of the very largest corporations in the economy. Thanks to the 
National Bureau of Economic Research and its collaborators, we now 
have good estimates of manpower and of income originating, by broad 
industrial groups, back to at least 1870. 75 It should be possible to dig out 

72 Lintner and Butters, "The Effect of Mergers . . . ," op. cit. 

73 Statistics of Income. During 1935-41 inclusive, balance sheets were filed on an 
unconsolidated basis, resulting in duplication and overstatement of the total of 
assets. The 1947 data, however, are largely consolidated. Thus the increase from 
1940 to 1947 is less than actually occurred. 

74 It would take the author too far afield to indicate why he favored the recent 
amendment to the Clayton Act: not because of but despite such arguments as those 
of the FTC. 

75 Simon Kuznets, National Product Since 1869 (New York, 1946) ; Studies in 
Income and Wealth, Vol. II (New York, 1949) . 


the employment and the value added or income generated by the hun- 
dred or so largest corporations starting about 1900 or perhaps even ear- 
lier, and to prepare decade estimates. Some of the work of identifying 
the largest concerns during the earlier decades has already been done 
by the Brookings Institute. The number of firms for which data were 
available would, of course, increase irregularly through time, but this is 
just where a measure of the type of c discussed above, would be most 
useful, since it is independent of the number of firms. 

Public Policy 

This article may be summarized in three statements. (1) The Ameri- 
can economy is highly concentrated. (2) Concentration is highly un- 
even. (3) The extent of concentration shows no tendency to grow, and 
it may possibly be declining. Any tendency either way, if it does exist, 
must be at the pace of a glacial drift. 

What are the implications for public policy? Strictly speaking, none. 
No deployment of facts, no analysis of them, however correct, can ever 
yield an imperative for action until those facts have been compared with 
some kind of standard or norm of desirability. For example, we might 
reason as follows: 

1. The American economy is highly concentrated. 

2. (a) It ought to be much less concentrated. 
(b) It ought to be much more concentrated. 

3. It gives few or no signs of becoming either more or less concen- 

4. Therefore, measures should be taken to 

(a) promote de-concentration; 

(b) promote concentration. 

Given the opposing premises in (2), the opposing conclusions flow 
with equal validity. Of course, statements of the type of (2) can them- 
selves be decomposed further into statements of the relation between 
concentration and some kinds of desirable and undesirable behavior, such 
as the relation between concentration and monopoly, however defined. 
This would take us very far from our present subject. But clearly the re- 
sults presented in this paper do not imply that public policy ought to be 
modified or, in particular, that antitrust policy should be changed. There 
is no escape from the pain of choice. 

But although a given set of facts is compatible with more than one pol- 
icy conclusion, this does not make it compatible with any conclusion, 
i.e., the facts are not simply irrelevant. Stated somewhat differently: our 
conclusions, if correct, should mean something for the atmosphere in 
which policy is formed. References to "the growth in economic concen- 
tration," or statements that "during the past 16 years | 1932-48] big busi- 
ness has been getting bigger and little business littler," or that "the forces 
of concentration [are] growing stronger by the hour," or that "concen- 


tration of industry is increasing so fast that no half-measures can stop it," 
and many others of like tenor — these must now be dismissed as un- 
founded. Concentration may be a problem, but for better or worse it is 
not threatening to engulf the economy. The moral, therefore, is simply 
that there is time to stop, look, and take thought. This may detract from 
the color and excitement of our public life, but it may also contribute to 
better results. 

Uses of Ideology 

It is unfortunately the case that much of the discussion and research in 
the field of big business and concentration is marked by emotion, and the 
procedure is often to choose one's side and be stuck with its story. 
There may be political repercussions from a research study, and the re- 
search worker who is interested in these results may hear the devil quot- 
ing Shakespeare to his purpose: 

Get thee glass eyes, 
And, like a scurvy politician, seem 
To see the things thou dost not. 

Nothing can be done about this; and nothing should be. That a man 
speaks with emotion, even prejudice, does not mean that he may not be 
speaking correctly. It is not who says it that counts but what he is saying. 
Nor is this all. The urge to convince others of the truth of one's "vision," 
as Joseph Schumpeter reminded the American Economic Association just 
a year before his death, "induces fact finding and analysis and these tend 
to destroy whatever will not stand their tests. ... And so — though we 
proceed slowly because of our ideologies, we might not proceed at all 
without them." 76 

76 Joseph A. Schumpeter, "Science and Ideology," American Economic Review, 
March, 1949, p. 359. 

Economies of Scale, Concentration, 

and the Condition of Entry in 

Twenty Manufacturing 



Ever since the merger movement of the late nineteenth century, Amer- 
ican economists have been recurrently interested in the extent to which 
large size is necessary for business efficiency. Was the merger move- 
ment necessary; was the rule of reason economically justifiable; can this 
or that concentrated industry be atomized without loss of efficiency? 
These continue to be important questions to students of recent industrial 
history and contemporary antitrust policy. In the last three decades, with 
the notion that plant or firm size is related to efficiency formalized in 
long-run average-cost or scale curves, there has been much speculation 
and some inquiry concerning the shapes and positions of those scale 
curves in various industries and the placement of existing plants and firms 
on them. 

To the economist qua economist, a knowledge for its own sake of the 
scale curves in particular industries is obviously unimportant. Only idle 
curiosity could justify his learning without further purpose how many 
barrels of cement a plant should produce to attain the lowest unit produc- 
tion cost, or how many passenger cars an automobile firm should make 
to minimize its production costs. But inferences which can be draw r n 
from such knowledge may be important in several ways. 

First, the proportion of the total output of its industry which a plant 
or a firm must supply in order to be reasonably efficient will determine 
the extent to which concentration in that industry is favored by the pur- 
suit of minimized production costs. In any industry, the minimal scales of 
plant and of firm which are required for lowest production costs — when 
these scales are expressed as percentages of the total scale or capacity of 
the industry and are taken together with the shapes of the scale curves 

* The American Economic Review, Vol. XI J V (1954), pp. 15-39. Reprinted by 
courtesy of the publisher and the author. 
t University of California. 



at smaller capacities — determine the degree of concentration by plants 
and firms needed for reasonable efficiency in the industry. 

Second, the same relation of productive efficiency to the proportion of 
the market supplied by a plant or firm in any industry will have a pro- 
found effect on potential competition, or on the disposition of new firms 
to enter the industry. If a plant or firm needs to supply only a negligible 
fraction of industry output to be reasonably efficient, economies of scale 
provide no deterrent to entry other than those of absolute capital re- 
quirements. If, however, a plant or firm must add significantly to industry 
output in order to be efficient, and will be relatively inefficient if it adds 
little, entry at efficient scale would lower industry selling prices or in- 
duce unfavorable reactions by established firms, whereas entry at much 
smaller scales would give the entrant a significant cost disadvantage. In 
this situation established firms can probably raise prices some amount 
above the competitive level without attracting entry. In general, the 
"condition of entry" — measured by the extent to which established firms 
can raise price above a competitive level without inducing further entry 
— becomes "more difficult" as the ratio of the output of the optimal firm 
to industry output increases. 1 

Third, the amount of money required for investment in an efficient 
plant or firm — as determined by size — will affect the availability of the 
capital necessary for new entry. When the supplies of both equity and 
loan capital in the range needed for a unit investment are either abso- 
lutely limited or positively related to the interest rate, the number of 
dollars required to establish an efficient plant or firm will clearly affect 
the condition of entry to an industry. 2 

Finally, a comparison of the scales of existing plants and firms in any 
/ industry with the most efficient scales will indicate whether plants and 
( firms are of efficient size, or whether or not the existing pattern of con- 
centration is consistent with reasonable efficiency. Have plant and firm 
concentration proceeded too far, farther than necessary, just far enough, 
or not far enough — from the standpoint of productive efficiency? A 
knowledge of scale curves is prerequisite to an answer. 

Although information on the relation of efficiency to scale thus has 
some importance, relatively little has been done to develop this knowl- 
edge through empirical research; economists have relied mainly upon a 
priori speculations and qualitative generalizations of the broadest sort. A 
popular American view is that economies of large-scale plant do exist — 
and that the efficiency of plants as large as are built may be conceded — 

1 See J. S. Bain, "Conditions of Entry and the Emergence of Monopoly," Monop- 
oly and Competition and Their Regulation, E. H. Chamberlin, ed. (London, 1954), 
for a development of this theory. 

2 But the absolute capital requirement for efficiency need not, as we move from 
one industry to another, be systematically related to the proportion of industry 
output needed for efficiency. 


but that further economies of large multiplant firms do not exist, of if 
they do, are strictly pecuniary in character and hence not to be sought or 
justified as a matter of social policy. 3 At the extreme it is argued that in- 
creasing the size of the firm beyond that of an efficient plant does not 
normally lower costs at all, so that the scale curve is approximately hori- 
zontal for some distance beyond this point. The dominant British view, 
expressed by such writers as Steindl, Florence, and E. A. G. Robinson, 
gives more credence to the alleged economies of large-scale firms. Both 
schools rely upon qualitative and substantially untested generalizations 
about productive and commercial techniques which supposedly deter- 
mine the response of production costs to variations in the scale of plant 
or firm. Yet in spite of the extremely sketchy nature of this sort of knowl- 
edge, it is common to presume, for instance, that there are numerous ex- 
amples of each of two sorts of oligopolistic industries — those where scale 
economies encourage a high concentration, and those where such econ- 
omies do not but something else does. 4 

Direct empirical investigation has not added much to our knowledge 
of scale curves. The principal studies employing accounting cost data are 
found in TNEC Monograph No. 13, and in later work by J. M. Blair, 5 of 
the Federal Trade Commission. Unfortunately the industries studied have 
been so few, the periods of time reviewed so remote and brief, and the 
use and interpretation of the statistical data in most instances so open to 
question that no reliable generalization regarding scale curves can be 
drawn from this body of material. There is more available in the way of 
profit-rate data for firms of various sizes, but here the unsupported as- 
sumptions which are normally necessary to argue from higher profits to 
lower costs are so numerous as to vitiate any attempt to infer scale curves 
from profit rates. Somewhat more satisfactory information has been de- 
veloped, for a very few industries only, through "engineering" estimates 
of the scale curve for plant or firm. But in general our information is 
such that we are ill-prepared to say much about actual scale curves and 
their implications. 


In the course of a recent general study of condition of entry to Amer- 
ican manufacturing industries, 6 it has been possible to develop some fur- 

3 See, e.g., TNEC Monograph No. 13, Relative Efficiency of Large, Medium- 
sized, and Small Business, pp. 95—139. 

It may be noted that the income-distribution effects of strictly pecuniary econo- 
mies may not be inconsequential in many settings. 

4 See, e.g., Fellner's "Case 1-a," "Case 1-b," and "Case 2" oligopolies, in his 
Competition Among the Few (New York, 1949), pp. 44 ff. 

5 See e.g., "Technology and Size," American Economic Review Proceedings, 
Vol. XXXVIII (May, 1948), pp. 121-52, and "Relation between Size and Efficiency 
in Business," Review of Economics and Statistics, Vol. XXIV (Aug., 1942), pp. 125—35. 

<; I wish to acknowledge the generous assistance provided for this study since 
1951 by the Merrill Foundation for the Advancement of Financial Knowledge, 


ther data on economies of scale therein. The portion of this information 
presented here concerns, for each of twenty selected manufacturing in- 
dustries: (1) the relationship of the output capacity of a plant of lowest- 
cost size to the output capacity of the industry, together with the shape 
of the plant scale curve at smaller sizes; (2) the relationship of the ca- 
pacity of a firm of lowest-cost size to industry capacity, and the firm 
scale curve at smaller capacities; and (3) the absolute amount of money 
capital required to establish an optimal plant and an optimal firm as of 
the current decade. 

These data have been developed almost entirely from managerial or 
"engineering" estimates supplied by certain firms in the industries in- 
volved; precisely, they reflect estimates of scale economies and capital 
requirements which were prepared, in response to detailed prearranged 
questioning, either by or at the direction of high-level executives in these 
firms. The general procedure for securing such data included: (1) a 
lengthy preliminary survey of each of the twenty industries, based on 
available monographs, documents, and other published and unpublished 
secondary materials; (2) the subsequent preparation for each industry of 
a separate, special, and rather lengthy series of questions designed to elicit 
certain information having bearing on the condition of entry; (3) secur- 
ing, after explaining the project involved and assuring confidentiality of 
replies, an advance offer of cooperation in answering these questions from 
executives in a large number of firms; (4) actual submission of the ques- 
tions, followed (except in those cases where cooperation was subse- 
quently withdrawn) by obtaining answers, in writing or orally or both. 
The method used thus involved neither shot-gun dissemination of an 
all-purpose questionnaire nor postprandial armchair quizzes, but rather a 
more or less hand-tooled questionnaire procedure in the case of each of 
twenty industries. 

The questions submitted relative to scale economies in each industry 
were designed in general to elicit information concerning the minimal 
plant size requisite for lowest unit costs and the shape of the plant scale 
curve at smaller sizes, the same information for the firm, and the capital 
required to establish a plant and a firm of most efficient size. Direct and 
(with exceptions to be noted below) explicit answers to these questions 
were normally secured. In many cases, there was abundant evidence in 
the length and documentation of replies of a careful estimating proce- 
dure; in some, figures submitted were frankly characterized as unsub- 
stantiated armchair guesses, though in most of these the respondents were 
very well qualified to guess. By and large, the writer is inclined to feel, 

through a grant made to the Research Group on the Monopoly Problem at Harvard 
University, directed by Dean E. S. Mason. Acknowledgment is also due for the 
assistance in preceding years of the Bureau of Business and Economic Research, 
University of California, Berkeley, where essential initial background studies were 


on the basis of checks against other sources and of comparisons of dif- 
ferent and independent replies to the same questions, that this is generally 
a fairly reliable body of data, in which the bulk of individual industry 
estimates are likely to be fairly accurate. The data have the advantage, so 
far as they are reliable, of reflecting "engineering" estimates in the sense 
that they represent expert ex ante predictions of the net relations of cost 
to scale, rather than an ex post comparison of gross cost results at dif- 
ferent achieved outputs. Thus they refer in general directly to scale 
curves as understood in economic theory. 7 

The twenty manufacturing industries studied may be designated as 
those producing cigarettes, soap, distilled liquor, shoes, canned fruits and 
vegetables, meat products, passenger automobiles, fountain pens, type- 
writers, flour, rubber tires and tubes, refined petroleum products, farm 
machinery, tractors, 8 steel, copper, cement, gypsum products, rayon, and 
metal containers. The sample was obviously not drawn at random. It was 
selected to obtain a maximum possible diversity of industry types con- 
sistent with the availability of data, but the fact that data have been more 
frequently developed for large and for highly concentrated industries 
than for others has resulted in some systematic differences between the 
sample and the whole population of manufacturing industries. 

The following characteristics of the sample deserve brief note: First, 
it features large industries, with fifteen of the twenty having value prod- 
ucts above a half billion in 1947. Whereas it includes only a little over 4 
per cent of the total number (452) of manufacturing industries in 1947, 
it accounts for about 20 per cent of the value product of all manufacture 
in 1947. 9 Second, it contains a substantially larger proportion of moder- 
ately and highly concentrated manufacturing industries than the total 
population. Nine industries of the sample had 75 per cent or more of 
value product controlled by four firms, three had 50 to 75 per cent so 
controlled, eight had from 25 to 50 per cent, and none less than 25 per 

7 The general time reference of all estimates is the period 1950 to 1952. From 
two to five such estimates were received in each of the twenty industries in ques- 
tion. Other sources of data which were available for some industries — such as com- 
parisons of accounting costs or the personal estimates of authors of industry studies 
— have been deliberately neglected here in order to give a more uniform consistency 
to the data presented. The only other data presented here, and these largely for 
expository purposes, are plant and firm concentration data prepared from the 1941 
Census of Manufactures. Since the engineering estimates which supply the bulk of 
our data were generally secured under guarantees of secrecy as to source, no ac- 
knowledgments or references to source can be supplied. 

8 For present purposes only we follow the Census in the dubious experiment of 
segregating tractors from other farm machinery. 

9 The total population of industries described, as well as all data on value prod- 
ucts and on concentration by firms, is derived (except as otherwise noted) from 
the 1941 Census of Manufactures, and in particular from a special analysis of con- 
centration prepared from this Census and published as an appendix in Hearings, 
Subcommittee on Study of Monopoly Power, Committee on Judiciary, H. R., 81st 
Cong., Serial 14, Part 2-B. 


cent controlled by four firms. 10 In the total population of manufacturing 
industries, the corresponding numbers in the four concentration classes 
were 47, 103, 164, and 138. This bias must be recognized in interpreting 

Otherwise, the sample is fairly representative. Eight industries are 
classed as making consumer goods, eight producer goods, and four goods 
bought by both producer and consumers. The outputs of eight are non- 
durable in use, whereas twelve are durable or semidurable. As to type of 
technique or process, five industries may be classified as engaged in 
processing farm products and four minerals, three as chemical industries, 
five as manufacturing or assembling mechanical devices, and three as in 
miscellaneous fabrication. 11 


Our first question concerns the shape and position of the plant scale 
curve (relating unit costs of production to the size of the individual fac- 
tory or plant) in each of the twenty industries, and the apparent con- 
sequences of economies of large plants for entry and for seller concen- 
tration. We are interested initially in the scale curve reflecting the relation 
of production cost to the output or capacity of the plant ivhen the latter 
are expressed as percentages of the total output or rated capacity supply- 
ing the market to be supplied by the plant. When output or capacity is 
expressed in these percentage terms, what is the lowest-cost or "optimal" 
size of plant and what is the shape of the plant scale curve at smaller 

An initial clue to the potential importance of economies of large plants 
is supplied by certain data on plant size assembled in the 1947 Census of 
Manufactures. This Census shows for each of many industries the num- 
ber of plants in each of several size-classes (size being measured by 
number of employees), and also the proportion of Census industry em- 
ployment and of total industry "value added" accounted for by each size- 
class of plants. From these data 12 certain inferences can be drawn about 
the sizes of existing plants. For exploratory purposes here I have tried to 

10 In three of the twenty cases, value added rather than value product figures were 
used by the Census in calculating concentration. For automobiles, registration rather 
than Census figures are followed in describing concentration, in both the sample 
and the total population, because of deficiencies in Census data. 

11 One further characteristic of the sample may be noted — Census industries have 
been selected which correspond fairly well to "theoretical" industries, or for which 
industry concentration as computed tends to reflect closely the relevant theoretical 
concentration of corresponding or component theoretical industries. This matter is 
discussed at length in J. S. Bain, "Relation of Profit Rate to Industry Concentration," 
Quarterly Journal of Economics, Vol. LXV (August, 1951), pp. 297-304. 

12 The data were previously used by the Federal Trade Commission for its study 
The Divergence between Plant and Company Concentration, 1941. The staff of the 
Commission has kindly made available its tabulated calculations on plant concen- 
tration as based on the Census data. 


develop from them some upper-limit estimates of the plant sizes requisite 
for greatest efficiency in the sample industries, by computing first the 
average size of plants in the largest size-class in each industry (expressed 
here as the percentage supplied per plant of the total value added of the 
Census industry), and second the maximum possible average size (sim- 
ilarly expressed) of the largest four plants in the industry. 13 If we neglect 
such obvious limitations as those of using value-added data, these esti- 
mates may be considered maximum percentages of the national industry 
outputs requisite for efficiency, on the grounds that in nearly every case 
we refer to the average size of a few of the largest plants actually built, 
and that the firms operating them were not restricted from building them 
to optimal scale. That is, they are generally multiplant firms which could 
bring a single plant to optimal scale before adding another, if indeed 
they did not in some cases duplicate optimal technical units on a single 

The results of these estimating procedures are as follows: Eighteen 
industries were examined (automobiles and copper being eliminated be- 
cause of gross deficiencies in Census data); for the eighteen the number 
of plants in the largest size-class lay between 3 and 15 in all but three 
cases; in those three it was large enough to make our estimates quite 
hazardous. The average share of Census industry value added supplied bv 
plants in the largest size-class ranged from 20.1 per cent (typewriters) to 
0.7 per cent (shoes), with a median at 3.8 per cent. The maximum possi- 
ble average share of the largest four plants ranged from 19.1 per cent 
(cigarettes) to 1.7 per cent (shoes), with a median at 7.9 per cent. 

The character of the data is more fully revealed in the frequency dis- 
tributions in Table 1. The first frequency column therein (fi) classifies 
industries according to the market-share interval within which the aver- 
age size of plants in the largest size-class of plants falls, market share be- 
ing measured by the percentage of the Census industry value-added sup- 
plied by a plant. The second frequency column (f 2 ) shows the same 
information when the plant size referred to in each industry is the maxi- 
mum possible average market share of the largest four plants. 

These findings, showing that in from seven to twelve of the 18 Census 
industries (depending on the method of estimate) the value added of the 
largest plants amounted to over 5 per cent apiece of total industry value 
added and that in from two to seven cases the figure was over 10 per 
cent apiece, suggest an importance for economies of large-scale plant 
which is substantial in some of these industries and small in others. But 
a detailed interpretation of the findings is not justified for several reasons. 
First, value added in a single year is a rather unsatisfactory measure of 

13 The latter figure is derived in general by attributing to all but the first four 
plants in the largest size class the minimum possible market share (i.e., for each the 
mean share of plants in the second size class) and by dividing the remainder of the 
total market share in the largest size class among the first four plants. 


"scale" as that term is ordinarily understood. Second, the largest plants 
as identified by the Census may have resulted from building multiples 
of optimal technical units on single locations, and if so, the figures pre- 
sented may overestimate optimal scales. Third, the data in question ex- 
press the output of the plant as a percentage of the total national value 
added within the Census industry, whereas in fact the theoretical indus- 
try or separate market which a plant supplies may be somewhat smaller. 14 
In these cases — where a Census industry is in fact made up of several 


Classification of Eighteen Census Industries According to 

Percentages of Industry Values-Added Supplied by 

the Largest Plants, 1947 a 

Percentage of Census 

Industry Value Added 

Supplied by the Average 

of the Largest Plants 

Number of Industries with the Largest Plant Size 
in the Specified Percentage Interval 

When "Largest Plant Size" 

Refers to Average Size of 

Plants in Largest Size 

Class of Plants 


When "Largest Plant Size" 
Refers to the Maximum Pos- 
sible Average Size of the 
Largest 4 Plants in Industry 

0- 2.4 
2.5- 4.9 
5.0- 7.4 






7.5- 9.9 



15.0-24.9 b 






a From 1947 Census of Manufactures. The composition of sample is described in the text. 
b The highest value in this class was 20.1 per cent. 

theoretical industries corresponding to distinct regional markets or prod- 
uct lines — the "percentage-of-industry-output" derived from Census data 
for large plants is very likely to be below the theoretically relevant fig- 
ure, 15 and revisions are in order. We thus turn at once to direct engineer- 
ing estimates of optimal plant sizes. 

Table 2 reviews the engineering estimates of the optimal scales of 
plants for twenty industry groups. In each case, the plant size referred 
to is the minimal physical production capacity of plant required for low- 
est production costs, this capacity being expressed as a percentage of 
total national capacity within the Census industry. In each case also the 
costs referred to are total production costs, including costs of outship- 
ment where the latter are strategic to the determination of optimal plant 

Table 3 summarizes the data of Table 2 by classifying industries ac- 

14 It may also conceivably be larger, as in the case where imports are omitted from 
Census data or where the Census industry is too narrowly defined, but these con- 
tingencies are not realized in any important degree in this sample. 

15 It will be if the plant specializes as to area or product line. 



cording to the market-share interval in which the mean estimated size of 
an optimal plant falls, when size is measured as a percentage of the na- 
tional industry capacity. These "engineering" data seem generally more 
satisfactory than those previously developed from Census figures. They 
reflect rational calculations rather than historical happenstance, and de- 
signed plant capacities rather than transient additions to value of output, 


Proportions of National Industry Capacity Contained in Single 

Plants of Most Efficient Scale, for 20 Industries, per 

Engineering Estimates circa 1951 

Percentage of 

Percentage of 

National Industry 

National Industry 


Capacity Con- 
tained in One 


Capacity Con- 
tained in One 

Plant of Minimal 

Plant of Minimal 

Efficient Scale 

Efficient Scale 

Flour milling 


Rubber tires and tubes 8 


Shoes a 


Rayon h 

4 to 6 

Canned fruits and 



Soap 5 

4 to 6 



Farm machines, ex tractors* 

4 to 6 

Distilled liquors b 

1M to \% 


5 to 6 

Petroleum refining" 


Automobiles k 

5 to 10 

Steel d 


Fountain pens 1 

5 to 10 

Metal containers 




Meat packing: 6 


10 to 15 


Ho to y 5 


2 to 2}4 

Gypsum products* 

2^ to 3 


10 to 30 

» Refers to shoes other than rubber. 

b Capacity refers to total excluding brandy. Costs refer explicitly to 4-year whiskey, packaged but ex tax. 

Optimal balanced integration of successive processes assumed. Outshipment largely by water assumed; optimal 
scale may be smaller with scattered market and land shipment. 

d Refers to fully integrated operation producing flat rolled products. 

e Percentages are of total nonfarm slaughter; diversified operation includes curing, processing, etc. 
f Combined plasterboard and plaster production assumed. 

e Purchase of materials at a constant price assumed; production of a wide variety of sizes assumed. 
h Refers to plant producing both yarn and fiber. 
* Includes household detergents, 
i Refers primarily to complex farm machines. 

k Plant includes integrated facilities for production of components as economical. Final assembly alone — 1 to 3 

1 Includes conventional pens and ballpoints. 
m Assumes electrolytic refining. 

although they still reflect percentages of the national capacities of Census 

It appears from them that in nine of the twenty industries an optimal 
plant would account for a quite small fraction of national capacity (under 
2% per cent), whereas in five others the fraction would run above 7% 
per cent. In general, the industries with slight economies of scale of plant 
are engaged in processing of agricultural or mineral materials, whereas 
greater plant economies are frequently encountered in industries making 
mechanical devices. The engineering estimates of the importance of 
economies of large plant present an over-all picture for these industries 


not greatly different from that derived by calculating average plant sizes 
in the largest plant-size intervals (column fi of Table 1), but they clearly 
ascribe less importance to such economies than the estimates of the max- 
imum possible average sizes of the largest four plants in each of these in- 
dustries (column f 2 of Table 1). 

Before we interpret these findings, however, two further matters must 
be discussed: the shapes of the plant scale curves at capacities short of 
the estimated optima, and the revisions in the estimates of optima which 
are needed if the division of Census industries into separate regions or 
product lines is recognized. 

As to the shapes of plant cost curves at capacities short of the estimated 


Classification of Twenty Industries According to Percentages 

of National Industry Capacities Contained in Single 

Plants of Most Efficient Scale 

(from Table 2) 

Number of Industries with Op- 
Percentage of National Industry timal Scale Plant (Per Mean 
Capacity Contained in a Plant Estimate) in the Specified Per- 
of Optimal Scale centage Interval (fa) 

0- 2.4 9 

2.5- 4.9 2 

5.0- 7.4 4 

7.5- 9.9 2 

10.0-14.9 2 

15.0-24.9 1 

Total 20 

optima, relatively fragmentary information has been received. In four 
industries the plant scale curve appears to be horizontal back to the 
smallest size considered, or % per cent of national industry output; these 
are flour, shoes, canned fruits and vegetables, and "fresh" meat packing. 
In ten cases — steel, metal containers, diversified meat packing, gypsum 
products, farm machinery, automobiles, fountain pens, copper, tractors, 
and typewriters — quantitative estimates of the shapes of the plant cost 
curves are not available, although in some cases (e.g., diversified meat 
packing and metal containers) it is suggested that substantially smaller 
than optimal plants would entail only slightly higher costs, whereas in 
some others (e.g., typewriters, automobiles, and tractors) a distinct rise 
in costs is suggested at half the optimal plant scale. For the seven re- 
maining industries, the estimated relation of production cost to plant 
scale is shown in Table 4, where costs of 100 represent the lowest at- 
tainable costs. 

A mixed picture again emerges. In some cases (liquor and cigarettes, 
for example) the rise of production costs at suboptimal scales is evidently 
quite small; in others (soap, petroleum refining, tires and tubes) it is 



moderate but by no means negligible; in some — e.g., rayon and cement 
— the rise is great. 16 One might hazard the guess that in from a half to 
two-thirds of all the industries sampled the upturn of the plant scale 
curve at suboptimal scales is such as to discourage very much smaller 
operations unless there are forces counterbalancing production cost dis- 
advantages. In the other one-third to a half of cases, a wide variety of 
plant sizes might prosper indefinitely in only slightly imperfect markets. 
The findings of Tables 2 and 3 however — reflecting as they do the 
percentages of national Census industry capacities supplied by single 
plants — can hardly be taken at face value so long as the suspicion remains 
that many Census industries may be broken into several separate and 
largely noncompeting regional or product submarkets and that a plant 
may specialize in only one such submarket. In these cases the relevant 

, TABLE 4 
Relation of Production Cost to Plant 
Scale in Seven Industries 


of National Industry Capacity 

in One Plant 











130 ' 

Distilled liquor 







Petroleum refining 







Tires and tubes 











Very high 






Above 105 





Above 102 

measure of plant size must be the proportion of the capacity supplying 
a submarket which is provided by an optimal plant, and this proportion 
will be larger than the proportion of national capacity provided by the 
same plant. 

In eleven of the twenty cases listed in Table 2, a revision of plant-size 
figures is in order because of the apparent division of the national market 
into distinct submarkets, coupled with plant specialization among them. 
In seven of these cases — flour, cement, petroleum refining, steel, metal 
containers, meat packing, and gypsum products — the important segmen- 
tation of markets is geographical in character; national markets are bro- 
ken into regions, and a single plant will mainly supply only one region. 
In the other four cases — shoes, canned fruits and vegetables, automobiles, 
and fountain pens — markets are divided to a significant extent among 
distinct product lines. In all cases, the relevant measure of plant size is 

1(5 It will be noted that the industries with the highest degrees of plant concen- 
tration are generally those on which it has been most difficult to secure quantitative 
estimates of the shape of the scale curve. In general, our information on plant scales 
seems sketchier and perhaps less reliable at this end of the sample. 


the percentage it may account for of the total capacity supplying any sub- 
market it may supply. 

The industries in which market segmentation is important are pre- 
dominantly those for which the percentages of national industry capaci- 
ties represented by single plants are quite small. The data for nine of the 
first ten industries in Table 2 require revision because of market segmen- 
tation, and only two for which revision is required lie in the range of 
high plant concentration nationally. Where technology does not give 
some importance to plant economies in industries of our sample, geog- 
raphy and product specialization (by plants) apparently do. Correspond- 
ingly, revised plant-size data showing percentages of individual submar- 
ket capacities will differ markedly from those in Tables 2 and 3. 

To make the revision mentioned, the optimal plant capacity for each 
of the eleven industries involved has been restated first as a percentage 
of the capacity supplying the largest submarket identified, and second as 
a percentage of capacity supplying the smallest of the major submarkets 
identified. For example, four major regional markets were identified in 
the petroleum refining industry. The proportion of national capacity 
supplied by a single optimal refinery had been estimated at 1 % per cent 
(Table 2); the corresponding percentages for the largest and smallest of 
the four major regional markets were 3% per cent and 11% per cent. In 
the fountain pen industry the proportion of aggregate national capacity 
supplied by an optimal plant was estimated at from 5 to 10 per cent. 
Dividing the market into high-price or gift pens and low-price pens in- 
cluding ballpoints (and recognizing differences in techniques for pro- 
ducing the two lines) the corresponding percentages become 25 to 33% 
per cent and 10 to 12% per cent. 

When these revisions have been made for the eleven industries, and the 
results combined with the unrevised data for the remaining nine, we are 
prepared to present two frequency distributions parallel to that in Table 
3 above. They classify industries according to the percentage of market 
capacity provided by an optimal plant, in the first case (column f 4 of 
Table 5) when the capacities of optimal plants in the eleven revised in- 
dustries are expressed as percentages of the total capacities supplying the 
largest submarkets in their industries, and in the second (column f 5 ) when 
optimal capacities in the eleven industries are expressed as percentages of 
the total capacities supplying the smallest major submarkets identified. 
The last column in Table 5 repeats column f 3 from Table 3 for purposes 
of comparison. 

Subjective judgments have inescapably influenced the content of col- 
umns f 4 and f 5 , particularly in the identification of regions, the decision 
as to what is a "major" region or product line, and the decision as to 
whether market segmentation is significant, but we have tried to follow 
available information and industry practice systematically. If there is a 
bias, it is in the direction of defining areas and product lines quite 


broadly, of considering only a few dominant areas for analysis, and of 
recognizing segmentation only if there is strong evidence supporting the 

Interpreting Table 5 with appropriate reference to the earlier discus- 
sion of the shapes of plant scale curves, we may emphasize the following 
conclusions about the importance of economies of large-scale plants 
within the industries of our sample. First, if the reference is to the largest 
submarkets of industries with segmented markets (plus the national 
markets of those with unsegmented markets), then in nine of the twenty 
cases an optimal plant would supply less than 5 per cent of its market, 


Classification of Twenty Industries' 1 by Percentages of Individual 

Market Capacities Contained in a Single Plant of 

Most Efficient Scale 

Number of Industries with Optimal Plai 

<n Scale in the 

Specified Percentage Interval 

Where Percentage 

Where Percentage 

Is That of the 

Is That of the 

Where Percentage 

Percentage of 

Total Capacity 

Total Capacity 

Is That of the 

Individual Market 

Supplying the 

Supplying the 

Total Capacity 

Capacity Contained 

Largest Recognized 

Smallest Recognized 

Supplying the 

in a Plant of 



National Market 

Optimal Scale 



(f 3 from Table 3) 

0- 2.4 




2.5- 4.9 




5.0- 7.4 




7.5- 9.9 























a The meat packing industry is considered for purposes of this table as only involving so-called 
fresh meat packing. 

and in five additional cases less than 7% per cent. If this is true and if, 
further, the plant-scale curve is usually fairly flat for a moderate range 
of suboptimal scales, then in many of these fourteen cases the scale re- 
quirements for an optimal plant should not provide a serious deterrent 
to entry. A firm constructing one reasonably efficient plant should not 
ordinarily induce serious repercussions from established firms in its mar- 

On the other hand in six cases — gypsum products, automobiles, type- 
writers, fountain pens, tractors, and copper — the proportion of the total 
capacity supplying either the national market or the largest submarket 
which is provided by a single optimal plant runs from 10 to 25 per cent. 
Precise data are largely lacking on the shapes of scale curves in these 


industries, but if they are much inclined upward at suboptimal scales 
(as is suggested qualitatively in several cases) then the economies of 
large plant should provide a very significant deterrent to entry to the 
markets in question. Further, a substantial degree of oligopolistic con- 
centration by firms might easily be justified by the pursuit of plant econ- 
omies alone. The substantial diversity of situations among industries of 
moderate to high concentration deserves considerable emphasis. 

The picture changes markedly if our attention shifts in the case of the 
eleven segmented industries from the largest to the smallest major sub- 
markets. Now we find that in eleven of the twenty cases (rather than six) 
the proportion of the relevant market capacity supplied by an optimal 
plant exceeds 10 per cent, and in six cases it exceeds 20 per cent. Plant 
economies sufficient to impede entry very seriously are potentially pres- 
ent in half or more of the cases, and high plant and firm concentration 
is encouraged by technology. The importance of plant economies thus 
potentially bulks large indeed in the smaller regional submarkets and the 
smaller product lines, whereas it is evidently less in the major submarkets 
and frequently so in the industries with relatively unsegmented national 


The extent to which further economies of large scale are realized if 
firms grow beyond the size of a single optimal plant has been a subject 
of controversy among economists. If a distinction is drawn between 
"production cost" and other advantages of scale — so that sales promo- 
tion, price-raising, and similar advantages of big firms are properly dis- 
tinguished from cost-savings in production and distribution — there is no 
general agreement among economists as to whether or to what extent 
the multiplant firm is more economical. 17 It thus may come as no surprise 
that business executives questioned on the same matter with regard to 
our sample of industries evidenced a similar diversity of mind. Very dis- 
tinct differences of opinion relative to the existence or importance of 
economies of multiplant firms were frequently encountered in the same 
industry, and in a pattern not satisfactorily explicable in general by the 
hypothesis that the individual would claim maximum economies for his 
own size of firm. Any findings presented here on estimates of economies 
of large-scale firm should thus be viewed as extremely tentative. 

Whatever the ostensible importance of economies of the multiplant 
firm, exploitation of them will not necessarily require the multiplant firm 
to control a larger proportion of any submarket than is needed for one 
optimal plant. In those instances where national markets are segmented 
regionally or by product lines, the multiplant firm may realize its econ- 

17 This disagreement is, as noted above, complicated further by difference of 
opinion as to whether the disputed economies are real or strictly pecuniary in 


omies while operating only one plant in each submarket. Then concen- 
tration by firms in individual submarkets is not further encouraged and 
entry is not further impeded 18 by economies of the multiplant firm. An 
optimal cement plant may supply about 1 per cent of national capacity, 
or percentages of regional capacity ranging very roughly from 5 to 30 
per cent in eleven regional submarkets. The fact that a multiplant cement 
firm could secure lower costs than a single-plant firm by operating one 
optimal plant in each of the eleven regions — thus accounting for 1 1 per 
cent of national capacity — would not imply that it need have a higher 
proportion of capacity in any one region than a single-plant firm of op- 
timal size. Except for an increase in absolute capital requirements, the as- 
sumed economies of the multiplant firm would not encourage regional 
market concentration or impede entry. 

Suppose on the other hand that there are economies of multiplant firms 
which are to be realized through operating two or more optimal-size 
plants either in a single submarket or in a single unsegmented national 
market. This will evidently encourage a concentration by firms in the 
relevant submarket or national market greater than that encouraged by 
plant economies alone, and will further impede entry. If a single plant 
of most efficient size would supply 5 per cent of the relatively unseg- 
mented national cigarette market, whereas a single firm operating three 
such plants could lower costs of production and distribution perceptibly, 
economies of the multiplant firm would favor greater effective concen- 
tration and provide further deterrents to entry to the cigarette industry. 

Findings relative to the economies of multiplant firms, together with 
certain related data, are presented in Table 6. The second column therein 
repeats the estimates of percentages of national Census industry capaci- 
ties required for optimal plants, from Table 2. The third column indi- 
cates the estimated extent of economies of multiplant firms (i.e., firms of 
sizes beyond those of single optimal plants), costs of distribution but 
not of sales promotion being included. The fourth column indicates the 
percentages of national industry capacities required for firms with low- 
est production plus distribution costs, while the final column shows the 
average percentage per firm of the national market supplied by the first 
four firms in 1947. The last provides a measure of actual concentration 
by firms. The estimates in question are entirely those of executives que- 
ried in connection with the investigation underlying this study. 

The data presented in Table 6 shed light on two questions: (1) to 
what extent do the economies of the multiplant firm tend to enhance 
concentration and impede entry, and (2) to what extent is the existing 
concentration by firms greater than required for exploitation of econo- 
mies of large plants and of large firms? 

Concerning the first question a varied picture appears. In eight indus- 

18 Except for the increase of absolute capital requirements. 



tries (Group 2 in Table 6) no definite estimate could be obtained of the 
extent, if any, of economies of the multiplant firm. This is in spite of the 
fact that in most of these industries the degree of concentration by firms 
substantially exceeds that requisite for exploitation of estimated econo- 
mies of the large plant. In six industries (Group 1 in Table 6) it was 


The Extent of Estimated Economies of Multiplant 

Firms in 20 Manufacturing Industries 


Percentage of 

Percentage of 





Share of the 

Industry Ca- 

Extent of 

Industry Ca- 


pacity Con- 


pacity Con- 


tained in One 

Economies (As a 

tained in One 

of First 4 


Percentage of 


Firms in 



Total Cost) 



Group 1: 

Canned fruits and vegetables 

Htoy 2 




Petroleum refining 





Meat packing: b 











Fountain pens 

5 to 10 








23.1 d 


10 to 30 




Group 2: 


Ho to ^ 

No estimate 



Distilled liquor 

IH to 1% 

No estimate 



Metal containers 

y 2 toi 

No estimate 



Tires and tubes 


No estimate 




4 to 6 

No estimate 



Farm machines, ex tractors 

4 to 6 

No estimate 




5 to 10 

No estimate 




10 to 15 

No estimate 



Group 3: 



Small, or 2 to 4 





Small, or 2 to 3 

2 to 10 



1 to 23/2 

2 to 5 

2 to 20 


Gypsum products 

2V 2 to 3 


27 to 33 



4 to 6 


8 to IS 



5 to 6 


15 to 20 


■ Market »hares are average percentages of 1947 national values of shipments unless otherwise indicated. 
b Plant percentages refer to total of nonfarm slaughter, firm percentages to wholesale fresh meat packing only. 

• Expresses average percentage of total value added rather than value of shipments. 

d Expresses average percentage of electrolytic plus other national copper refining capacity, 1947. 

• Expresses approximate average percentage of total 1951 passenger car registrations. 

the consensus that economies of the scale of firm beyond the size of a 
single optimal plant were either negligible or totally absent. In these cases 
estimated cost savings of the multiplant firm cannot justify concentration 
beyond that required by plant economies alone (either in submarkets or 
in unsegmented national markets) nor can they make entry any more 
difficult than it is already made by plant economies. With respect to the 


first four industries in the group, a multiplant firm with plants in several 
regions or product lines would, according to the estimates received, re- 
alize no net cost savings by virtue of this aspect of its organization. In 
the second three industries in this group, however, economies of the large 
plant alone are sufficient to support a high degree of concentration by 
firms and to impede entry. 

In the remaining six industries (Group 3 in Table 6) perceptible econ- 
omies were attributed to the multiplant firm. The extent of these econo- 
mies is in no case huge, being characterized as slight or small in three 
cases and as in the two to five per cent range in the remaining three. 
Nevertheless, two or three percentage points on total cost can be signifi- 
cant in any industry if the ratio of operating profits to sales is not be- 
yond five or ten per cent and if product differentiation and other market 
imperfections are not dominant. What further tendency toward concen- 
tration and what further impediment to entry would the existence of 
these economies imply? 

The optimal multiplant firm as estimated in Group 3 of Table 6 in- 
cludes two or three optimal plants in the soap industry, three or four 
in the cigarette industry, four or five in the shoe industry, and about ten 
in the gypsum products industry. Estimates for the steel and cement in- 
dustries run all the way from one or two to ten plants per optimal firm, 
and the range of disagreement among authorities is wide. Applying these 
estimates, the proportion of national industry capacity needed for best 
efficiency in a multiplant firm is raised; but is the proportion of the ca- 
pacity supplying any particular regional or product submarket also 
raised? It will not be if the efficient multiplant firm includes only one op- 
timal plant per submarket, and it will be if it includes two or more per 
submarket or if the national market is unsegmented. 

In Group 3 in Table 6 no more than one optimal plant per region is 
attributed to the optimal firm in cement or in steel, and the proportion 
of any regional market which need be supplied for efficiency is thus not 
increased by the incidence of economies of the multiplant firms. In the 
remaining four cases the conclusion is different. Soap and cigarettes have 
relatively unsegmented national markets, and the proportion of the mar- 
ket required for best efficiency is doubled, trebled, or quadrupled by the 
emergence of economies of the multiplant firm. In shoes the assumed 
specialization to a single product line of the four or five plants needed 
for efficiency raises the requisite firm concentration by product lines by 
corresponding multiples. In the gypsum industry it was evidently as- 
sumed that an optimal firm would operate several plants in each of one 
or more major regions. In all of the last four cases, therefore, economies 
of the multiplant firm encourage greater effective concentration by firms 
and impede entry. But in these cases (possibly excepting shoes) the econ- 
omies of the large firm were characterized as slight, so that the effects 
just listed may be weak. 


With respect to the effect of the economies of multiplant firms on con- 
centration and on entry, these conclusions appear. In eight of twenty in- 
dustries in our sample, no estimate was obtained of the extent of these 
economies. In two-thirds of the remaining cases, economies of the mul- 
tiplant firm were held either to be absent, or to take such a form that 
exploitation of them would not require higher proportions of market 
control by the firm in any submarket. In one-third of the remaining 
cases, some encouragement to higher concentration by firms in submar- 
kets was provided, but it was a small encouragement in view of the gen- 
erally slight economies attributed to the large firm. Economies of the 
large-scale firm apparently do not represent a major force encouraging 
concentration or deterring entry in this sample of industries. The data on 
which this guess rests, however, are far from adequate. 

Our second question concerns the extent to which the existing degree 
of concentration by firms within industries is justified by the estimated 
economies of large plants and firms. This is a rather complicated ques- 
tion, and may be broken down into three subquestions: (1) Is the exist- 
ing concentration by firms for national Census industries justified by the 
economies of single large-scale plants? (2) If not, is the existing concen- 
tration by firms nevertheless consistent with no higher concentration 
within individual submarkets than is required by a single efficient plant 
— i.e., need there be more than one optimal plant per large firm in any 
one submarket? (3) In any case, to what extent is the multiplant charac- 
ter of large firms apparently justified by the economies of such firms? 

A first approximation to answers to these questions may be made by 
taking the concentration figure in Column 5 of Table 6 as a simple 
and crude measure of national industry concentration by firms. 19 On the 
basis of this measure, the answer to the first subquestion is simple and 
unsurprising — concentration by firms is in every case but one greater 
than required by single-plant economies, and in more than half of the 
cases very substantially greater. Generally it is only within some of the 
industries with very important economies of large plant — e.g., fountain 
pens, copper, typewriters, autos, tractors, farm machines — that concen- 
tration by firms has not been much greater than required by single-plant 
economies. Even in these cases it may be two or three times as great 
as thus required. In the other cases concentration by firms tends to be a 
substantial or large multiple of that required by single-plant economies. 
Remembering that we are dealing in general in this sample with the more 
concentrated industries, it might be said in summary that nearly all of 
the industries tended to become moderately or highly concentrated (by 
firms) whether economies of the single plant were important or not. 

The second subquestion is whether the existing degree of concentra- 

19 The average share of national industry output per firm for the first four firms 
obviously is smaller than the market share for the first firm, larger than that for the 
fourth firm, etc. 


tion by firms is consistent or inconsistent with the existence of a single 
optimal plant per firm in each recognized submarket. In seven of the nine 
cases where the national market has been considered substantially unseg- 
mented — copper, typewriters, liquor, tires and tubes, rayon, farm ma- 
chines, tractors, soap, and cigarettes — the degree of concentration by 
firms within a single market is greater than required by such plant econ- 
omies, although in all but two of the seven cases (liquor and tires and 
tubes) it is greater by at most a multiple of three or four. This last 
is found probably in part because economies of the large plant seem very 
important in most of these industries. 

In eight of the remaining eleven cases — canned goods, petroleum re- 
fining, meat packing, fountain pens, metal containers, cement, steel, and 
gypsum products — the degree of national concentration by firms is not 
grossly inconsistent with the larger firms on the average having but a 
single optimal plant per submarket in each of several submarkets. (This 
is certainly not to deny that the largest single firms may have more than 
this and probably do; we refer only to the average of the largest four 

In the last three cases — flour, automobiles, and shoes — the degree of 
concentration by firms exceeds by a multiple of two or three that re- 
quired for each of the four largest firms on the average to have an opti- 
mal plant in each submarket. In general, our showing is that in ten of 
twenty industries the existing degree of concentration by firms, as meas- 
ured by the average size of the largest four firms, is significantly greater 
than required for these firms to have only one optimal plant per submar- 
ket; in the other ten cases concentration is at least roughly consistent 
with such a condition. 

The third subquestion concerns the extent to which the existing de- 
gree of concentration by firms is justified by the exploitation of econ- 
omies of multiplant firms. We will go no further with this question here 
than a comparison of the fourth and fifth columns of Table 6 will take 
us. In Group 1 in that table, the alleged absence of any economies of 
multiplant firm implies that there is no justification in terms of costs 
for the excess of concentration by firms over that required for single 
efficient plants, although in one case (typewriters) the existence of an 
excess is uncertain, and in four others (all but copper) it is not neces- 
sarily accompanied by accentuated concentration in individual sub- 
markets. Here, therefore, the lack of an evident cost justification for 
multiplant firms raises not so much the issue of concentration in separate 
markets as the issue of the other advantages and disadvantages of a 
diversified firm operating in each of several related submarkets. 

In Group 2 no estimates of multiplant economies are available; we need 
say no more than that in five of eight cases (excluding metal containers, 
farm machines, and tractors) there is a concentration by firms much 
greater than that required for efficient plants in each submarket, and that 


this requires evaluation from a cost standpoint. In only one of the indus- 
tries in Group 3 (shoes) does the degree of concentration by firms seem 
to have clearly exceeded that required for economies of production and 
distribution by the large firm. 

In the sample as a whole the existing degree of concentration by multi- 
plant firms lacks a clear cost justification in perhaps thirteen of twenty 
cases, although in seven of these we have a simple lack of any definite 
estimates. In two more cases the multiplant phenomenon is not very im- 
portant. Further information is needed on this matter, particularly with 
reference to cases in which multiplant firm organization has increased 
effective concentration in individual submarkets or in unsegmented na- 
tional markets. 


The effect of scale economies on the condition of entry so far em- 
phasized is transmitted through their influence on the share of market 
output which an efficient plant or firm will supply. This impact is im- 
portant, but it is not proportional to the importance of scale economies 
measured in such terms as the absolute number of employees or the ab- 
solute size of investment required for an optimal plant or firm. This is be- 
cause the proportion of a market supplied by an optimal plant or firm 
(which determines the degree of oligopolistic interdependence between 
the potential entrant and established firms) depends not only on the ab- 
solute size of the plant or firm but also on the size of the market. Thus 
an investment of over $200 million dollars might add only one per cent 
to national steel capacity, whereas an investment of $6 million might add 
five or ten per cent to the capacity for producing fountain pens. In addi- 
tion to the effect of scale economies on entry via the proportion of the 
market an efficient entrant will supply, there is a distinct and not closely 
correlated effect via the absolute size of the efficient plant or firm, or, to 
choose a popular measure, via the total money investment needed to es- 
tablish such a plant or firm. 

To determine the importance of scale economies in establishing suffi- 
cient capital requirements to impede entry seriously, we have queried 
the same sources on the investment requisite for the most efficient plant 
or firm in the twenty industries sampled. The findings relative to capital 
requirements for the large plant are fairly comprehensive, and are sum- 
marized in Table 7. Column 2 of this table shows the estimated percent- 
age of national industry capacity provided by one efficient plant, and 
Column 3 the total investment required to establish such a plant (ordi- 
narily including working capital) as of about 1951. The industries are 
grouped according to the importance of scale economies from the previ- 
ously emphasized percentage standpoint. The first category of industries 
are those in which a single efficient plant will supply no more than 5 per 
cent of the largest submarket or unsegmented national market; the sec- 



Estimated Absolute Capital Requirements for Plants of Estimated 

Most Efficient Scale, circa 1951, for 20 Industries 

Percentage of 

National Industry 

Capacity Provided by 

One Efficient 

Plant (from 

Total Capital Required for One 


Table 2) 

Efficient Plant a 

Category 1 : 

Flour milling 


$700,000 to $3,500,000 



$500,000 to $2,000,000 

Canned fruits and vegetables 

l Atoy 2 

$2,500,000 to $3,000,000 



$20,000,000 to $25,000,000 

Distilled liquor 

IX to \H 

$30,000,000 to $42,000,000 

Petroleum refining 


$193,000,000 ex transport facilities 
$225,000,000-$250,000,000 with 
transport facilities 

Meat packing b 


Very small 

2 to 23^ 

$10,000,000 to $20,000,000 

Tires and tubes 


$25,000,000 to $30,000,000 

Category 2: 



$265,000,000 to $665,000,000 d 

Metal containers 6 

y 2 to3 

$5,000,000 to $20,000,000 


4 to 6 

$50,000,000 to $75,000,OOO e 
$90,000,000 to $135,000,OOO f 


4 to 6 

$13,000,000 to $20,000,000« 

Farm machines ex tractors 

4 to 6 

No estimate 


5 to 6 

$125,000,000 to $150,000,000 

Category 3: 

Gypsum products 11 

2V 2 to 3 

$5,000,000 to $6,000,000 


5 to 10 

$250,000,000 to $500,000,000 

Fountain pens 

5 to 10 

Around $6,000,000 



No estimate 


10 to 15 

Around $125,000,000 


10 to 30 

No estimate 

a These estimates generally exclude anticipated "shakedown losses" of new entrants, which in some cases may 
be large and prolonged. 

b The two rows of estimates refer alternatively to fresh and diversified meat packing. 

Percentage of an efficient plant in the largest regional market may exceed 5 per cent. 
d Excludes any investment in ore or coal. 

e Acetate rayon. 

1 Viscose rayon. 

8 Excludes working capital. 

h Percentage of an efficient plant in the largest regional market may exceed 10 per cent. 

ond includes those where the corresponding percentage is 5 to 10 per 
cent; the third includes those where the percentage is above 10 per cent. 
We may thus observe the extent to which the "percentage effect" of scale 
economies is of the same order as their "absolute capital requirement ef- 

The findings in Table 7 speak fairly clearly for themselves, but a few 
comments may be in order. First, there is no evident correlation of the 
absolute capital requirements for an efficient plant with the percentage 


of market output supplied by it. The size of the market is an erratic var- 
iable forestalling such a correlation. Second, absolute capital require- 
ments for an efficient plant in all the manufacturing industries examined 
are large enough to restrict seriously the ranks of potential entrants; even 
500,000 dollars, the smallest amount listed, will not be forthcoming from 
savings out of salary or from the winnings in a poker game. 

Third, the absolute capital requirements in some cases reinforce but 
in other cases weaken the "percentage effect" on entry of economies of 
scale of plant. For each of the eight industries in Category 1 in Table 7, 
for example, the percentage of market output supplied by a single plant 
seems small enough to provide no serious deterrent to entry. In three 
of these cases — flour milling, shoes, and canned goods 20 — the absolute 
capital requirements are also so small that entry may not be seriously 
restrained thereby. But in four others, capital requirements ranging from 
10 to 42 million dollars per plant provide a greater deterrent, and in one 
(petroleum refining) they impose a truly formidable barrier. 

In the six industries of Category 2, where the "percentage effect" on 
entry of economies of scale of plant is moderate, it is strongly reinforced 
in four cases (possibly excepting metal containers, and farm machines, 
for which there is no estimate) by absolute capital requirements. The ef- 
fect is very much increased in both the steel and cigarette industries. In 
the six industries of Category 3, where the "percentage effect" appears 
quite important, it is strongly reinforced in the cases of automobiles and 
tractors by absolute capital requirements, but in the fountain pen and 
gypsum industries capital requirements are relatively small. Thus a gen- 
erally mixed picture regarding the dual effects of economies of large 
plant emerges. 

The extent to which economies of multiplant firms as already noted 
increase the capital requirements for efficiency may be readily ascer- 
tained by comparing the findings of Table 6 with those of Table 7. 
Since the existence of such economies was denied in six industries, not 
estimated in eight others, and held to be slight in at least half the re- 
maining six, detailed comment on this matter does not seem justified. 


When the answer provided by empirical investigation to an initial in- 
quiry concerning the values of certain economic data is that the values 
are highly irregular and variegated, and when the answer is therefore 
found only in a great array of numbers, any brief summarization of the 
findings may be difficult to make and misleading if attempted. Since this 
situation is encountered with respect to each of the major questions posed 
at the beginning of this paper, no comprehensive summary of findings 
will be attempted here. Certain salient conclusions may be restated 

As well as in fresh meat packing. 


briefly, however, in each case with the proviso that they may have gen- 
eral validity only so far as the sample of industries selected is generally 
representative of moderately to highly concentrated manufacturing in- 
dustries in the United States. 

Regarding the importance of economies of large plants, the percentage 
of a market supplied by one efficient plant in some cases is and in some 
cases is not sufficient to account for high firm concentration or to im- 
pede entry. Where it is, these economies might easily propagate high 
concentration and serious impediments to entry; the number of cases 
where it is sufficient increases as we refer to the smaller regional or prod- 
uct submarkets in various industries. A significant corollary of these find- 
ings is that the following popular horseback observations are apparently 
not true: that economies of scale of plant are never or almost never im^ 
portant in encouraging oligopoly or impeding entry, and that such econ- 
omies always or almost always are important in these ways. The picture 
is not extreme in either direction and not simple. 

The economies of large plants frequently erect formidable barriers to 
entry in the shape of absolute capital requirements. Moderately to very 
high barriers of this sort were found in all but four or five of the in- 
dustries studied. The height of such barriers is not clearly correlated 
with percentage of the market supplied by a single plant, so that a rela- 
tively independent influence on entry is discovered. 

The economies of large multiplant firms are left in doubt by this in- 
vestigation. In half the cases in which definite estimates were received, 
such economies were felt to be negligible or absent, whereas in most of 
the remainder of cases they seemed slight or small. Perhaps the fre- 
quently expressed suspicion that such economies generally are unimpor- 
tant after all is supported, and perhaps we are justified in saying that we 
have had difficulty in accumulating convincing support for the proposi- 
tion that in many industries production or distribution economies of large 
firms seriously encourage concentration or discourage entry. 

Our reference here has of course been strictly to the effect of the size 
of the plant or firm on the cost of production and distribution, and 
thereby on entry and on concentration. Needless to say, parallel studies 
of other factors bearing on entry, including the effects of scale on price 
and on sales promotion, are required for a full evaluation of the entry 

Monopoly and Oligopoly by 


The growth of individual firms to great size through merger with ri- 
vals is an outstanding development of modern economic history. As late 
as 1890, Marshall could view the life history of the firm as a silhouette 
of that of man in an age of high infant mortality: the firm began as a 
small venture; if it survived the early years, it straggled along or grew 
at a rate governed by the entrepreneur's ability — occasionally reaching 
large size if his ability was extraordinary or his children's abilities great — 
but eventually it languished into obscurity and then into oblivion. 1 The 
whole process usually took place, one infers, in one or two generations. 
I have no reason to question the realism of this picture in the age of 
noncorporate enterprise; and there are reasons for not wholly abandon- 
ing it even today. An anthropomorphic theory of the growth of the firm, 
however, scarcely fits our modern giants. There are no large American 
companies that have not grown somewhat by merger, and probably very 
few that have grown much by the alternative method of internal expan- 
sion. 2 

The present paper seeks to summarize some of the major episodes in 
the development of the merger movement, with special reference to the 
question of monopoly. The discussion is restricted to so-called "horizon- 
tal" combinations, which are quantitatively much the most important 
form of merger. 3 

* The American Economic Review, Vol. XL (Proceedings of the American Eco- 
nomic Association, 1950), pp. 23-34. Reprinted by courtesy of the publisher and the 

t Columbia University. 

1 Principles of Economics (8th ed.; London, 1920), Bk. IV, chaps, xi, xii. 

2 Unless otherwise indicated, size of the firm is to be measured relative to the 
size of the industry. 

3 In 1937, 85.7 per cent of the manufacturing establishments belonging to "central 
offices" (i.e., multiple-plant firms) were engaged in "uniform" activities. Of course 
many of these plants were constructed by the parent concern, and there are other 
deficiencies in the data, but it is probable that horizontal mergers are more im- 
portant than all other forms of interplant relationship combined (see TNEC Mono- 
graph No. 27, The Structure of Industry, p. 164). 





We wish to examine the conditions under which it is profitable for 
competing firms to merge for monopoly. It is expedient to begin with 
four unpromising assumptions, all of which will be relaxed or defended 
subsequently: (1) long-run average and marginal cost of production are 
equal for firms of all relevant sizes;* (2) entry of new firms is free, al- 
though not necessarily inexpensive; (3) the demand for the output of 
the industry is stable; (4) the specialized resources ("fixed factors") em- 
ployed in the industry are indestructible. 

Under these conditions, will mergers for monopoly occur? The tempt- 
ing offhand reply is in the negative, because under these conditions there 
can be no monopoly profits in the long run; the first two conditions are 
sufficient to insure this. This offhand reply, however, is to the question: 







C A 

Figure 1 

will mergers for monopoly exist? It is not an answer to our question: 
will mergers for monopoly occur? They may occur. 

The argument that monopolies may be profitable even under these un- 
favorable conditions will be developed with a partial geometrical illustra- 
tion. Consider an industry meeting the four conditions listed above and 
consisting of numerous identical firms which are in long-run competitive 

4 This assumption will be discussed below, but perhaps a remark is called for on 
the indeterminacy of the output of the firm under competition when its long-run 
average cost curve is horizontal. The simplest way to eliminate the indeterminacy 
is to sacrifice the perfection of competition (but nothing else) by having each 
firm have a demand curve with an elasticity of (say) —100. 


equilibrium. Each firm will have the short-run cost curves displayed in 
Figure 1, and it will be operating at output OA, price OB, and making 
no profits. All the firms are now merged into a monopoly, and each 
plant (= former firm) now has a pro rata share of aggregate demand, 
AR, with corresponding marginal revenue, MR. Accordingly it operates 
at output OC and makes profits of OC times DE. Entry of new firms 
therefore takes place, and the pro rata demand curve of each plant in the 
merger now shifts to the left, price falls, and profits diminish. 5 Even- 
tually the number of rivals will grow until the merger is reduced to the 
long-run equilibrium level of permanent loss, since neither the merger 
nor the new rivals can withdraw from the industry. 

The simple but important conclusion to be drawn from this argument 
is that a merger for monopoly may be profitable, in the sense that the 
present value of the monopoly profits and (so to speak) monopoly losses 
is positive. If the entry of new firms is not too rapid, the merger may 
make monopoly profits for a considerable period; and, even though 
thereafter the losses are permanent, their discounted value need not be 
so large as to wipe out the initial gains. The essence of the explanation 
of mergers under these conditions therefore lies in the time required to 
achieve long-run equilibrium; and this essence lingers in the more gen- 
eral case. 

If we relax our assumptions "2," "3," and "4," the prospects of net gain 
from merger for monopoly are increased in frequently encountered cir- 
cumstances. If the specialized resources of the merger are not inde- 
structible, investment can be withdrawn from the industry so that, after 
the initial period of gain and a subsequent period of loss, the long-run 
equilibrium will be attained, with the merger receiving a competitive 
rate of return on its investment in the industry. 6 If the industry's de- 
mand is growing, the amount of resources the merger must withdraw 
will be reduced; and, if the demand is growing sufficiently rapidly, no in- 
vestment need be withdrawn: the merger can maintain its absolute size 
but decline in relative size. 7 If the entry of new firms and the expansion 
of rivals can be hindered or prevented, of course the monopoly profits 
will accrue for a longer period. If the rate of entry is a function of price 
and profits, the merger can reduce or retard entry by a lower price pol- 

5 The explicit analysis can be carried through by a conventional application of 
the dominant-firm analysis; that is, by constructing the demand curve for the mo- 
nopoly by subtracting from the aggregate quantity demanded at each price the 
amount that the new firms (acting competitively) will sell (see my Theory of Price 
[New York, 1946], p. 227). 

6 The period of loss arises because in general it requires less time to increase 
than to withdraw investment. 

7 This assumes (with what validity I do not know) that the rate of entry of 
new firms will not be increased by the existence of the merger; but see the last 
point in the paragraph. 


icy; in effect it buys a longer period of monopoly at the price of a lower 
rate of monopoly profits. 

Let us consider now the mechanics of mergers for monopoly. (We de- 
fer the question of why mergers occur when they do.) If there are rela- 
tively few firms in the industry, the major difficulty in forming a merger 
is that it is more profitable to be outside a merger than to be a par- 
ticipant. The outsider sells at the same price but at the much larger out- 
put at which marginal cost equals price. Hence the promoter of a 
merger is likely to receive much encouragement from each firm — almost 
every encouragement, in fact, except participation. In order to overcome 
this difficulty, it will often be necessary to make the participation of each 
firm contingent on that of other firms and execute the merger in a single 
act. We know too little of the theory of coalitions to be able to predict 
the percentage of the industry that will be merged, but of course it must 
be fairly high if it is to have any purpose. 

If there are relatively many firms in the industry, no one firm plays 
an important role in the formation of the merger; and it is possible for 
the merger to expand in a more gradual process and acquire firms on less 
exacting terms. In fact, several firms may enter upon programs of growth 
by merger. 

Let us return to our first two assumptions. Our first assumption — that 
there are neither economies nor diseconomies of scale — will please few 
beside Euler. Two widely accepted, and somewhat inconsistent, beliefs 
clash with this assumption: (1) mergers are effected to obtain the econo- 
mies of large-scale production and (2) the diseconomies of scale are the 
chief bulwark of competition. Both these beliefs will be discussed below; 
here we shall enter briefly into discussion of the validity of the assump- 
tion of constant returns to scale. 

The comparative private costs of firms of various sizes can be meas- 
ured in only one way: by ascertaining whether firms of the various sizes 
are able to survive in the industry. Survival is the only test of a firm's 
ability to cope with all the problems: buying inputs, soothing laborers, 
finding customers, introducing new products and techniques, coping 
with fluctuations, evading regulations, etc. A cross-sectional study of the 
costs of inputs per unit of output in a given period measures only one 
facet of the firm's efficiency and yields no conclusion on efficiency in 
the large. 8 Conversely, if a firm of a given size survives, we may infer 
that its costs are equal to those of other sizes of firm, being neither less 
(or firms of this size would grow in number relative to the industry) nor 

8 As commonly conducted, statistical comparisons of costs or rates of return are 
not even conclusive on the "static" problem. They demand arbitrary asset valuations 
to avoid the tautological result that differences in costs measure returns on differences 
in capital values; and they usually cover too short a period to avoid the regression 
problem (on which, see M. Friedman and S. Kuznets, Incomes from Independent 
Professional Practice [New York, 1945], chap. vii). 


more (or firms of this size would decline in number relative to the in- 

A combination of this argument and casual observation suggests that 
the economies of scale are unimportant over a wide range of sizes in 
most American industries, for we commonly find both small and large 
firms persisting. We shall recur to this matter, but two observations 
should be made now. The first is that our analysis of mergers still holds 
if there are minor economies or diseconomies of scale, but fails if they 
are large. With large diseconomies, mergers are unprofitable; with large 
economies, monopoly or oligopoly is inevitable, and there will not be 
many rivals to merge. The second point is that the equality of private 
costs carries no implication that social costs of firms of different sizes 
are equal. 9 

Free entry — our second assumption — may be defined as the condition 
that long-run costs of new firms if they enter the industry will be equal 
to those of firms already in the industry. This does not mean, as many 
infer, that a new firm can enter and immediately be as profitable as an 
established firm. We do not begrudge the new firm a decent interval in 
which to build its factory; we should be equally willing to concede a 
period during which production is put on a smooth-running schedule, 
trade connections are developed, labor is recruited and trained, and the 
like. These costs of building up a going business are legitimate invest- 
ment expenses, and, unless historical changes take place in the market, 
they must be equal for both established and new firms. 10 

With this understanding, free entry seems a valid characterization of 
most American industries. One may concede this and still argue that, be- 
cause of the large capital requirements necessary to establish a new com- 
pany of minimum efficient size, free entry is often difficult, and firms in 
industries with (absolutely) large capital requirements have a sheltered 
position. I have as little basis for my skepticism of this argument as its 
many adherents have given for supporting it. 

This brief discussion leaves many questions about mergers unan- 
swered. We shall attempt to answer some of them — and in the process 
discover new questions — by an examination of the merger movement in 
America. We shall find it useful to divide this history into two periods, 
in which monopoly and oligopoly, respectively, were the primary goals. 

9 A comparison of the social costs of firms of different sizes would require, for 
example, the elimination of differences in private costs arising out of differences in 
"bargaining power" in purchasing inputs. It is tempting to argue that if the large 
firm is not more efficient than the small firm in private terms, it is less efficient in 
social terms (for then its monopolistic advantages are eliminated). I am inclined 
to yield to the temptation, although small firms have some private advantages (chain- 
store taxes) and large firms have private disadvantages (maintaining good public 
relations) . 

10 On this view, the infant-industry argument for tariffs is mistaken (at least when 
external economies do not enter) . 



The era of merger for monopoly ended in this country roughly in 
1904, when the Northern Securities decision made it clear that this ave- 
nue to monopoly was also closed by the antitrust laws. The transition 
was abrupt in a historical sense. It is revealed by the fact that the 
United States Steel Corporation, which had quietly picked up in 1902-4 
a few steel firms overlooked in the haste of organization, felt it necessary 
to obtain permission from President Theodore Roosevelt to acquire the 
Tennessee Coal and Iron Company in 1907. 11 

When and why did the merger movement begin? Sporadic mergers, 
often founded on marriage, are no doubt as ancient as man; probably for 
long they were occasional and relatively small in scale, and they were 
offset by the divestitures necessary to endow sons in a more fertile age. 
In this country mergers for monopoly began on a large scale only in the 
eighties, they reached a minor peak at the beginning of the nineties, and 
they attained their pinnacle at the end of the century. 12 

Our theory is that mergers for monopoly are profitable under easy as- 
sumptions that were surely fulfilled in many industries well before the 
mergers occurred. The only persuasive reason I have found for their late 
occurrence is the development of the modern corporation and the mod- 
ern capital market. In a regime of individual proprietorships and partner- 
ships, the capital requirements were a major obstacle to buying up the 
firms in an industry, and unlimited liability was a major obstacle to the 
formation of partnerships. 

General incorporation laws antedate the Civil War, 13 but the powers of 
these early corporations were severely limited. They could not hold stock 
in other corporations; they could not merge with another corporation; 
limits were placed on their capitalization; often they could not do busi- 
ness outside the state of incorporation; exchange of capital assets for 

"Union Steel (1902), Troy Steel Products (1903), and Clairton Steel (1904) to- 
gether had twice the ingot capacity of Tennessee Coal and Iron. On the last merger, 
see the hearings of the Stanley Committee, Hearings before the Committee [of the 
House] on Investigation of United States Steel Corporation (Washington, 1911), 
Parts 1-6. 

12 The number of combinations with capitalization of 1 million dollars or more, 
as compiled by Luther Conant, varied as follows: 





























(Eliot Jones, The Trust Problem in the United States [New York, 1921], p. 39). 
There was an earlier era of railroad consolidations (see, e.g., George P. Baker, 
The For?nation of the New England Railroad Systems [Cambridge, 1937], especially 
chap. xi). 

13 See G. H. Evans, Business Incorporations in the United States, 1800-1943 
(New York, 1948) . 


stock required the unanimous consent of the stockholders; etc. Only in 
the eighties did New Jersey initiate the competition among states for 
corporations, which in twenty years eliminated almost every restriction 
on mergers. 14 In this same period the New York Stock Exchange devel- 
oped into an effective market for industrial securities. These institutional 
changes seem to be the proximate causes for the development of the 
merger movement in the last two decades of the nineteenth century. 

Almost invariably the leading firms joined together simultaneously, as 
our theory leads us to expect. 15 The combinations frequently attained 
high percentages of national output but seldom became strict monopo- 
lies. The contemporary estimates of their shares of the market are rough, 
and little attention was paid by the estimators to the shares of the firm 
in particular geographical and product markets. With these provisos, we 
may note that the mean share of the market controlled by the mergers 
studied by the Industrial Commission was 71 per cent. 16 In the ninety- 
two large mergers studied by Moody, the distribution by share of mar- 
ket was similar: seventy-eight controlled 50 per cent or more of the out- 
put of the industry; fifty-seven controlled 60 per cent or more; and 
twenty-six controlled 80 per cent or more. 17 Even in Dewing's fourteen 
industries in which the mergers failed, the mean percentage was 54. 18 

Almost invariably the share of the merger in the market declined 
substantially as time went on. Sometimes the entry of new firms was 
successfully prevented or delayed by ruthless warfare (National Cash 
Register), patents (Eastman, United Shoe Machinery), or coercion of 
suppliers or buyers (American Tobacco). These instances are not nu- 
merous, however, and such tactics were successful chiefly in small indus- 
tries; in steel, sugar refining, agricultural implements, leather, rubber, 
distilleries, cans, etc., the dominant company lost ground relative to the 
industry. 19 

Why was merger preferred to collusion? Part of the answer lies in the 

14 See E. Q. Keasbey, "New Jersey and the Great Corporations," Harvard Law 
Review, 1899-1900, pp. 198-212, 264-78; W. C. Noyes, A Treatise on the Law of 
Intercorporate Relations (Boston, 1902), and, especially, R. C. Larcom, The Delaware 
Corporation (Baltimore, 1937). 

15 The most prominent exception, Standard Oil, is in one sense no exception. It 
is more an instance of involuntary merger, and its history seems to me one of the 
relatively few cases appropriately analyzed in analogy to warfare. 

16 The distribution was: 

Per Cent Companies 

25-50 1 

50-75 11 

75-100 10 

(United States Industrial Commission, Report, Vol. XIII, passim.) 

17 John Moody, The Truth about the Trusts (New York, 1904) , p. 487. 

18 A. S. Dewing, Corporate Promotions and Reorganizations (Cambridge, 1914), 
p. 526. 

19 For some instances see my Five Lectures on Economic Problems (London, 
1949) , Lecture 5. 


prima facie illegality of collusion after 1890. This point should not be 
pressed, however. The effectiveness of the Sherman Law in dealing with 
conspiracies was not clear until 1899, when the Addyston Pipe case was 
decided; 20 and there was a contemporaneous wave of amalgamations in 
England, where conspiracies were unenforcible but not actionable. 21 
Mention should also be made of the conflicting tendencies of the greater 
durability of mergers and the ability to avoid diseconomies of scale 
through collusion. I am inclined to place considerable weight upon one 
other advantage of merger: it permitted a capitalization of prospective 
monopoly profits and a distribution of a portion of these capitalized prof- 
its to the professional promoter. The merger enabled a Morgan or a 
Moore to enter a new and lucrative industry: the production of monopo- 

It is sobering to reflect on the attitudes of professional economists of 
the period toward the merger movement. Economists as wise as Taussig, 
as incisive as Fisher, as fond of competition as Clark and Fetter, insisted 
upon discussing the movement largely or exclusively in terms of indus- 
trial evolution and the economies of scale. They found no difficulty in 
treating the unregulated corporation as a natural phenomenon, nor were 
they bothered that the economies of scale should spring forth suddenly 
and simultaneously in an enormous variety of industries — and yet pass 
over the minor firms that characteristically persisted and indeed flour- 
ished in these industries. One must regretfully record that in this period 
Ida Tarbell and Henry Demarest Lloyd did more than the American 
Economic Association to foster the policy of competition. 


One great change has taken place in the merger movement since the 
Northern Securities decision: the share of the industry merged into one 
firm has fallen sharply. In the early period, as we have seen, the leading 
firm seldom merged less than 50 per cent of the industry's output; in the 
later period the percentage has hardly ever risen this high. The new goal 
of mergers is oligopoly. 

The change has been most striking in the industries which were 
merged for monopoly at the beginning of the century. The merger firm 
has declined continuously and substantially relative to the industry in al- 
most every case. The dominant firm did not embark on a new program 
of merger to regain its monopolistic position, however; the new mergers 
were undertaken by firms of the second class. The industry was trans- 
formed from near-monopoly to oligopoly. Cement, cans, petroleum, au- 
tomobiles, agricultural implements, and glass are examples. We may il- 

20 See W. H. Taft, The Anti-Trust Act and the Supreme Court, and J. D. Clark, 
The Federal Trust Policy. 

21 J. H. Clapham, An Economic History of Modern Britain (Cambridge, 1938), 
Vol. Ill, chap. iv. 


lustrate the development by the steel industry (Table 1 ) : United States 
Steel's share of ingot production dropped sharply, but the company ab- 
sorbed only two small rivals (Columbia Steel, 1930; Geneva plant, 1945), 
and the chief mergers have been Bethlehem and Republic. Even if one 
lumps together the (say) four largest firms in the industry, in general 
there has been a decline in the concentration of production. 22 

The merger movement has also reached the many-firm industries in 
the later period. We may measure mergers between 1919 and 1937 in a 


Mergers by Leading Steel Firms Measured by Percentage 
of Industry Ingot Capacity 

Per C 

:nt of In- 



Per Cent of 

Acquired by 

Industry's Capacity 













U.S. Steel 




















Jones and Laughlin 











Youngstown Sheet 










American Rolling 
































Source: Compiled from directories of iron and steel works. The data presented by United States Steel Corpora- 
tion indicate a much larger control in the first decade, presumably because the industry's capacity was overstated 
in the directories (see TNEC, Hearings, Part 26, pp. 13, 852). 

rough fashion by comparing the number of manufacturing establish- 
ments belonging to central offices at the two dates (Table 2); it appears 
that the food industry was the chief center of merger activity, although 
the paper and printing and iron and steel industries also saw much merger 
activity. National Dairy is perhaps the most striking example of merger 
in the food industries — it acquired 331 firms in the decade ending in 
193 3 23 — but Borden, General Foods, General Mills, and the bakery chains 
also date from this period. 24 In general, such mergers led to local oligop- 
oly in the primary products (fluid milk, bread, etc.) and to national oli- 
gopoly in lesser products such as cheese. 

22 See my Five Lectures, op. cit., pp. 63 ff. 

23 Federal Trade Commission, Agricultural Income Inquiry (Washington, 1938), 
Vol. I, p. 237. 

24 There was also much merging in fuel and ice; City Ice and Fuel is perhaps 
the largest merger, but several others, such as American Ice and Atlantic Company, 
were very active. 


The Sherman Law seems to have been the fundamental cause for the 
shift from merger for monopoly to merger for oligopoly. Sometimes its 
workings were obvious, as when Standard Oil was dismembered and 
when the leading baking mergers were prevented from combining. 25 
More often, however, its workings have been more subtle: the ghost of 
Senator Sherman is an ex officio member of the board of directors of 
every large company. This explanation for the new direction of mergers 
is vulnerable to the criticisms that it is simple and obvious, but no plau- 
sible alternative explanation is available. 26 

It is my impression — based chiefly upon the more modest issuance of 


Manufacturing Establishments in Central Offices, 
1919 and 1937 




Food and kindred products 



Textiles and their products 



Iron and steel and their products 



Lumber and its remanufactures 



Leather and its finished products 



Paper and printing 



Liquors and beverages 



Chemicals and allied products 



Stone, clay, and glass products 



Metal and metal products other than iron and steel 



Tobacco manufactures 



Vehicles for land transportation 



Miscellaneous industries 






Source: 1919 data from W. L. Thorp, The Integration of Industrial Operations (Washington, 
1924), p. 113; 1937 data, which are only roughly comparable, from TNEC Monograph No. 27, The 
Structure of Industry, p. 211. 

securities of mergers and the apparent ease of entry into the new merg- 
ing industries — that the mergers for oligopoly in the later period have 
been less effective in restraining or postponing competition than the ear- 
lier mergers for monopoly. This is not to argue, however, that they left 
competition as they found it; indeed, the one important weakness in the 
Sherman Act as it is sometimes interpreted is the belief that oligopoly 
affords a satisfactory form of organization of our economy. This belief 
is apparently held, as it was certainly fostered, by one of the greatest of 
contemporary judges, Learned Hand, the author of the famous dictum 
that control by one firm of 64 per cent of an industry may not be mo- 

25 Agricultural Income Inquiry, op. cit., p. 308. 

26 It is suggestive that mergers for monopoly continued to be typical in England 
in the twenties (see Patrick Fitzgerald, Industrial Combination in England [Lon- 
don, 19271). 


nopoly and that 33 per cent surely is not. 27 It is true, no doubt, that oli- 
gopoly is a weaker form of monopolization than the single firm, but it 
is not so weak a form that it can be left to its own devices. If this 
view — which is almost universally held by modern economists — is cor- 
rect, then our chief task in the field of antitrust policy is to demonstrate 
beyond judicial doubt the social undesirability of permitting oligopoly 
by merger (or by other methods) in large American industries. 


The foregoing survey of the merger movement raises a set of interre- 
lated questions; they concern the economies of scale, the capital market, 
and the entry of firms into an industry. 

The broad sweep of our discussion would suggest that the private dis- 
economies of large-scale production are only an occasional and minor 
barrier to merger for monopoly. The chief barriers to monopoly, in ad- 
dition to the Sherman Act, have been the capital requirements of mergers 
and the tendency of rivals to grow in number and size. 

To find in an imperfect capital market a bulwark of competition seems 
paradoxical, but the paradox is not deep. Until recent times the personal 
distribution of wealth set a limit upon the size of firms, and modern eco- 
nomic societies have been sufficiently egalitarian to make personal mo- 
nopolization of large industries impossible. The corporation and the se- 
curities markets have severed the connection between personal wealth 
and industrial size and thus weakened the institutional basis of competi- 
tive enterprise. 

The diseconomies of scale offer a weak supplement to the limitations 
once provided by personal wealth. Properly interpreted, conventional 
theory does not contradict this tentative finding. We customarily find in 
entrepreneurship the limitation to the size of firm, and we find the chief 
tasks of the entrepreneur arising out of uncertainty. Much, although of 
course not all, uncertainty stems from the competitive behavior of rivals, 
so that entrepreneurship may well be subject to increasing returns to 
relative size as well as to decreasing returns to absolute size, with no 
clear verdict for either force over a wide range of sizes. 

We are thus led to "new entry" as the chief defense of competition — a 
most unseemly reversion to the ruling economic theory of 1900. It is now 
popular to deprecate the importance of new entry because few firms can 
accumulate the capital necessary to produce efficiently in the great in- 
dustries. To the extent that the criticism rests on the alleged economies 
of scale, I have argued that it is mistaken; to the extent that it rests on 
imperfections of the capital market it runs contrary to (but is not nec- 
essarily inconsistent with) the argument we have advanced that the cap- 

27 ". . . it is doubtful whether sixty or sixty-four per cent would be enough [to 
constitute monopoly]; and certainly thirty-three per cent is not" (United States v. 
Aluminum Co. of America, 148 F. [2d] 424). 


ital market has been improving — in certain directions, at least — too 
much! Yet there is support for the skeptics of easy entry in the fact 
that the mergers for monopoly have frequently been very profitable. 

Such inconclusive conclusions are not too troublesome. This paper is 
designed to be only an introduction to the merger problem. To this end, 
it is sufficient if I emphasize again the significance of the movement. To 
the theorist it offers a stimulating challenge: the merger movement does 
not fit too well into the received categories of stable competition and 
irresistible monopoly. To the student of social policy it offers the prom- 
ising hypothesis: it is possible to change the trend of industrial organiza- 
tion by the lackadaisical enforcement of an antitrust law. And to the 
student of social sciences it offers the supremely optimistic — and pessi- 
mistic — suggestion: when economists agree that a movement is inevi- 
table, it is not. 


An Alternative Approach to the 

Concept of Workable 



Economists, recognizing the shortcomings of the theory of perfect 
competition in framing public policy for oligopolistic markets, recently 
have endeavored to define a more realistic standard of economic per- 
formance — workable or effective competition. 1 Since the concept owes 
its creation to a public policy need and not to the logic of abstract the- 
ory, it can, at best, be divorced only in part from value judgments. To 
the unswerving socialist, nothing short of complete government owner- 
ship and control of all economic activity may make for a workable so- 
ciety. To him, no economy is workably competitive if it is privately 
competitive. Yet a proponent of unlimited free enterprise would prob- 
ably view any economy not subject to governmental intervention as 
workably competitive. Thorough-going antitrust sympathizers may re- 
ject any definition of workable competition that falls noticeably short of 
pure competition. On the other hand, economists who subscribe to the 
Schumpeterian theory of creative destruction may willingly accept, in 
fact, insist upon the acceptance of, far less. Nor is it quite clear that the 
degree of competition clamored for by constituents of each school of 
thought is always compatible with the economic framework of the so- 
ciety they hope to create or maintain. Those who insist that workable 
competition falls little short of perfect competition, in their efforts to 
preserve a free enterprise economy, may be unwittingly leading us to- 
ward stringent state controls. For, it is difficult to visualize the mainte- 
nance of atomized industries by any means other than strong police 
power. On the other hand, passive acceptance of a high degree of con- 

* The American Econo?nic Review, Vol. XL (1950), pp. 349-61. Reprinted by 
courtesy of the publisher and the author. 

The paper has profited considerably from criticisms made by Professors George W. 
Stocking and Roland McKean of the Vanderbilt University economics department. 

t Princeton University. 

x The term "workable competition" was first used by J. M. Clark in his article 
"Toward a Concept of Workable Competition," American Economic Review, Vol. 
XXX (June, 1940) , pp. 241-56. 



centration in many industries on the theory that other entrepreneurs with 
other products will destroy it through a natural evolutionary process may 
be an invitation to industrial fascism. 

That the term "workable competition" defies precise definition, even 
in the presence of agreed upon standards of workability in some political 
or moral sense, becomes evident when one attempts to define the "indus- 
try" to which the concept is to be applied. Concentration ratios are often 
employed to indicate the degree of monopoly which exists in a particular 
industry, but concentration ratios in themselves might be quite meaning- 
less unless they relate to a well-defined market. For example, E. I. du 
Pont de Nemours and Co., Inc. is the sole producer of a particular syn- 
thetic yarn known as nylon; but du Pont accounts for only about 10 
per cent of total synthetic yarn production and accounts for only an 
infinitesimally small percentage of total domestic yarn production of all 
kinds. Obviously, therefore, any measure of du Pont's monopoly power 
in the textile industry is dependent upon whether one has in mind the 
domestic nylon market, the synthetic yarn market, or simply the yarn 

Most definitions of workable competition have been patterned largely 
after the traditional definition of perfect competition. That is to say, the 
definitions consist of listing a set of conditions the fulfillment of which 
determines whether or not a specific industry is to be judged as work- 
ably competitive. By and large, none of these definitions presents insu- 
perable difficulties so long as their application is limited to near-perfectly 
competitive industries and well-defined market areas; i.e., although few 
would argue that the American cotton textile industry has traditionally 
fulfilled all the conditions of perfect competition, it is unlikely that any 
would argue that competition in cotton textiles has not been effective or 
workable. Most of our industries, however, are not so highly competi- 
tive that they differ but little from purely competitive industries nor are 
market boundaries in our economy always clearly delineated. When ap- 
plied to monopolistically competitive and oligopolistic markets, the 
weaknesses of most definitions of workable competition begin to appear. 
Yet, these are the market areas with which public policy is largely con- 
cerned. This paper purports (1) to set forth the actual conditions that 
exist in a fairly typical oligopolistic industry — the rayon yarn producing 
industry; (2) to demonstrate the difficulties that arise when an attempt 
is made to determine the workable competitiveness of an industry by ap- 
plying to it a specific set of conditions; and (3) to suggest an alternative 
approach to the problem of defining workable competition that may pos- 
sibly be more useful for purposes of public policy. 


Apart from the fact that the author has made a rather extended study 
of the industry, the rayon yarn industry offers itself as a particularly 


good candidate upon which the various definitions of workable compe- 
tition may be tested. Since the industry is composed of a limited num- 
ber of producers, it falls far short of fulfilling one of the most essential 
conditions of pure competition. On the other hand, market practices in 
the industry have never precipitated an antitrust case. 2 Hence, at the out- 
set, there is no basis for a priori judgment or public policy bias. In ad- 
dition to these features, however, the industry poses many of the debat- 
able questions that would necessarily arise in the process of determining 
whether or not an industry is workably competitive. The significant 
characteristics of the domestic rayon yarn industry are as follows: 3 

1. Rayon is produced in the United States under the viscose, acetate 
and the cupramonium processes. 4 All basic patent rights to these proc- 
esses have become public property. The American Viscose Corporation 
monopolized the industry until 1920, at which time its American rights 
to the Cross and Bevans and Topham patents expired. Since 1920, ap- 
proximately twenty new firms have entered into rayon production under 
one of the three processes listed above. A number of small firms did not 
survive the 1930-1933 depression. Concentration ratios have declined 
rapidly since 1920. Currently, the largest producer accounts for 33 per 
cent of total domestic output and the four largest producers account for 
about 75 per cent. 

2. The rayon industry is an oligopoly nominally consisting of four- 
teen firms. If cognizance is taken of inter-company affiliations, however, 
the actual number of financially independent producers is eleven. The 
chief limitation to the number of firms is an economic one: The long- 
run cost curve for the firm, at least to the extent it is indicated by avail- 
able plant cost curves and other data, places the rayon industry among 
the natural oligopolies. Economies of scale obtained through rapid ex- 
pansion of capacity by entrenched firms have made it increasingly diffi- 
cult for new firms to enter the industry since 1930 in spite of the grad- 
ual dissemination of technical "know how" and the expiration of several 
special patents that occurred over the past twenty years/ However, two 

2 The Federal Trade Commission investigated charges against the ten largest 
viscose yarn producers (1934-37) that they conspired to fix prices between October, 
1931 and May, 1932. Charges were made by a large textile fabricator. The Federal 
Trade Commission issued a cease and desist order in 1937 but mentioned no specific 
price-fixing practices. Viscose Co., et al.; Docket No. 2161; Summary of Decisions 
of the Federal Trade Commission. 

3 These characteristics have been taken from a larger study by the author en- 
titled Price and Output Behavior in the Domestic Rayon Industry. The original 
study was presented as a Ph.D. thesis at Harvard University, September, 1948. It 
is currently being revised for publication. Although, for purposes of this paper, the 
features of the rayon industry could as well be hypothetical as real, documentation 
of them may be found in the thesis deposited in the Harvard University Library. 

4 Although there are important chemical differences among yarns produced under 
the three processes, rayon yarn will be treated as a homogeneous commodity in this 


new rayon producers, Beunit Mills, Inc., and the United Rayon Corpora- 
tion, have entered the field since the War and were planning to have 
plants in operation before the end of 1949. 

3. The industry has been characterized by a very rapid rate of expan- 
sion of production facilities, 5 but only during the period between 1920 
and 1930 was any of this increase due to the entrance of new firms. A 
sharp secular decline in price (12.8 per cent per year) and rate of re- 
turn on investment (100 per cent 1915-19 to 6.9 per cent 1930-38) has 
accompanied the rising trend in investment and output. Therefore, 
trends in prices, profits and investment are consistent with those which 
would have been expected under highly competitive conditions. 

4. The rayon price series indicates a high degree of short-run stability 
relative to rayon sales and to natural fiber yarn prices. Prices often re- 
main unchanged for from one to two years. However, over the course 
of several business cycles rayon price has moved in almost perfect con- 
sonance with silk and cotton yarn prices. Both synthetic and natural fiber 
yarn producers view the yarn market as highly competitive. There are 
several logical explanations for the short-run stability of rayon price: 
(1) The prices for all variable cost factors (wages, chemicals, wood pulp, 
etc.) are extremely stable. (2) All rayon producers are large corpora- 
tions; it is generally conceded that this form of business organization 
transmits price changes to the market much more sluggishly than do 
small-owner controlled corporations and proprietorships. (3) Rayon 
producers seem to feel that a stable yarn price stimulates long-run de- 
mand; it eliminates possible inventory losses and the necessity of hedg- 
ing by textile fabricators. (4) Few producers initiate price changes inde- 
pendently; most producers usually wait for a large producer to initiate 
a price change (see 5, below). 

5. The price relationship among rayon producers has varied some- 
what with respect to particular phases of the business cycle, but there 
is fairly conclusive evidence of price leadership. The largest rayon pro- 
ducer has been the first to announce a list price change in all but a few 
instances; the two largest producers have accounted for 90 per cent of 
all list-price revisions since 1926. However, in periods of severe depres- 
sion such as 1930-32 and, to a less extent, 1934 and 1938, many list-price 
changes merely took de jure recognition of de facto market prices. 
There is tremendous pressure upon producers, particularly smaller pro- 
ducers, to maintain close to maximum capacity output for cost considera- 
tions. Short-run plant cost curves have a steep negative slope that con- 
tinues all the way to installed capacity output. The smaller the plant, the 
steeper the slope over the range of 75 per cent to 100 per cent of in- 
stalled capacity. Thus, when the rayon market becomes severely de- 

5 Many statistics textbooks use the rayon production series to illustrate the ap- 
plicability of the Gompertz curve. See F. C. Mills, Statistical Methods (rev. ed.; 
New York, 1938), pp. 671-75. 


pressed, small producers will shade prices in order to maintain a high 
rate of output rather than abide by list prices and commit financial sui- 
cide. Also, extremely tight rayon markets have occasionally given rise to 
premiums above list price of varying magnitudes and, since the War, 
there has been some variation in quoted prices among producers for the 
first time in the history of the industry. Market conditions intermediate 
to an acute yarn shortage and a severe depression, however, appear to be 
characterized by close agreement between list price and net realized 
price and by identical quoted prices for all producers. 

6. Similar to the price series, the rayon output series is highly stable 
in the short-run relative to the deliveries series. Generally, irregular and 
seasonal fluctuations in demand are absorbed by inventory fluctuations. 
Several factors account for the reluctance on the part of rayon producers 
to curtail output to meet temporary market disequilibria: (1) Rayon is 
produced by a continuous process; plant shut-downs are costly and sev- 
eral weeks are usually required to get plants back up to top operational 
efficiency. (2) As indicated above, total unit costs increase rapidly as 
output is contracted below capacity. This is true when all variable factors 
may be adjusted to the new production schedule; the increase in unit 
costs is aggravated in the immediate short-run that would probably 
include seasonal and irregular fluctuations in demand for which complete 
adjustment of variable factors is seldom if ever made. 

7. By way of an appraisal of the above points, the arbitrary power that 
rayon producers may exercise over their price is extremely limited if we 
concede that producers attempt to maximize profits. In no case has out- 
put curtailment in the industry proceeded by as much as 25 per cent of 
installed capacity without an accompanying price reduction. A conflu- 
ence of several economic forces accounts for this price breaking point: 
(1) Large firms have reasons for suspecting off -list selling when only 75 
per cent of installed capacity operations is necessary to fill orders at 
existing list prices. (2) Total unit costs (under conditions of stable fac- 
tor costs) are considerably higher when firms operate at 75 per cent of 
installed capacity than when they operate at 100 per cent. (3) The prices 
of the competitive natural fibers have probably been noticeably reduced 
by the time textile fabricators are demanding only 75 per cent of capacity 
rayon output; hence, it behooves the nominal price-leader or some other 
firm to reduce rayon prices in order to regain part of its market lost to 
natural fibers and thereby maintain a higher level of output. 

8. Although rayon producers do not seem to be willing to engage in 
much price competition when the market is "normal," they vigorously 
compete among themselves in other ways. The rate of technological in- 
novation in the rayon industry has been almost unbelievably high. For 
example, output per man-hour increased about tenfold between 1925 and 
1938. Also, numerous acid recovery systems have been devised to cut 
costs. In their efforts to produce at a cost lower than their rivals, rayon 


producers have engaged in vigorous competition. Further, rayon pro- 
ducers have aggressively explored and exploited the long-run demand 
possibilities for their product. The quality of rayon has been steadily im- 
proved or altered so that it could reach new fabric markets. In 1920 
rayon was, at best, a cheap silk substitute. Today rayon is used in the 
production of all kinds of wearing apparel and household and industrial 

9. The domestic rayon yarn industry has received a considerable 
amount of tariff protection. 6 Since the passage of the Tariff Act of 1930, 
domestic producers have been relatively free from foreign competition. 
Under the recent Geneva Trade Agreement, however, import duties on 
rayon yarn were reduced by 50 per cent on January 1, 1948. Although 
it is too early to tell what the long-run effects of tariff reduction upon 
the domestic yarn market will be, rayon imports increased considerably 
in 1948. 

Obviously, these are not all of the characteristics of the domestic rayon 
yarn industry that must be determined before a final verdict can be 
rendered upon the workable competitiveness of the industry. However, 
for illustrative purposes, they serve as a tolerably good testing ground for 
the applicability of some of the more familiar definitions of workable or 
effective competition. They at least serve our purpose sufficiently well to 
bring into sharp relief the extent to which one must engage in making 
value judgments in any such process. 


There have been several serious attempts made to define the term 
"workable competition." In no case, however, has an author set forth 
conditions so completely devoid of value judgments or so all-embracing 
that he feels free to acclaim the universal applicability of his definition. 
Professor J. M. Clark, the creator of the term, places much emphasis 
upon rivalry among selling units and the "free option of the buyer to buy 
from a rival seller or sellers of what we think of as 'the same' product." 7 
Professor Clark concedes, however, that the "specific character of com- 
petition in any given case depends on a surprisingly large number of con- 
ditions — so many, in fact, that the number of mathematically possible 
combinations runs into the hundreds of thousands — and suggests the 
possibility that every industry may be in some significant respect differ- 
ent from every other, or from itself at some other stage of develop- 

6 When discussing the workable competitiveness of the domestic rayon yarn 
market in this paper, the tariff policy will be accepted as a datum. Public policy can 
always be directed toward making any domestic market more or less competitive 
than it now is by eliminating tariffs, subsidizing imports, imposing higher tariff 
walls, etc. 

7 John M. Clark, op. cit., p. 243. 

8 Ibid. 


On a number of counts, the domestic rayon industry would be worka- 
bly competitive under Professor Clark's definition. He himself would 
probably place the rayon industry under the following caption in his 
classification of competition scheme: 9 

II. Modified, intermediate or hybrid competition. 

A. Standard Products, few producers. The most important cases involve 
formally free entry, but no exit without loss. 
3. Quoted prices, sloping individual demand curves. 

b. Demand schedules indefinite (limited freight absorption). 

There is room for little doubt about the existence of rivalry among 
the producers of rayon. In depressed markets rivalry has taken the form 
of price cutting and price shading. Even during normal market periods 
when the price leader's list price seems to have prevailed in the industry 
there has been intense rivalry among producers in the form of cost 
cutting and quality competition. Except possibly for the brief interval 
between October, 1931, and May, 1932, when the price-fixing conspiracy 
was alleged, it can be fairly concluded that rayon purchasers have had a 
real option before them as to their source of supply. At times when all 
producers rigidly adhered to the price leader's quoted price, the buyer's 
option was limited to quality and grading differences among producers. 
However, to the extent buyers regard rayon, nylon, silk, cotton and 
wool yarns as "the same" commodity, they have always had, in addition 
to the above, a real price option. Hence, on most counts the rayon yarn 
industry would qualify as workably competitive under Professor Clark's 
definition. At least, any disagreement with this conclusion would evolve 
from a different interpretation of such terms as "rivalry," "option," and 
"the same" commodity. 

To Professor Stigler, "An industry is workably competitive when 
(1) there are a considerable number of firms selling closely related prod- 
ucts in each important market area, (2) these firms are not in collusion, 
and (3) the long-run average cost curve for a new firm is not materially 
higher than that for an established firm." 10 The workable competitiveness 
of the rayon industry under Professor Stigler's difinition would depend 
upon the interpretation given to the term "closely related products." If 
all textile yarns are in some sense to be considered as closely related, and 
the competition among them in specific market areas indicates that they 
should be so considered, the rayon industry (as a small segment of the 
total yarn market) would be considered as highly competitive under the 
above definition. If, however, the product of one rayon producer be 
considered as closely related only to other rayon, other features of the 
definition come into play and the conclusion is not so evident. The work- 
able competitiveness of the rayon industry would then depend upon how 

9 John M. Clark, op. cit., p. 245. 

10 George J. Stigler, "Extent and Bases of Monopoly," American Economic Re- 
view, Vol. XXXII, Suppl. (June, 1942), pp. 2-3. 


Professor Stigler would answer the following questions: Do fourteen 
producers constitute "a considerable number of firms"? Do sporadic 
price shading and price cutting so invalidate the normal acceptance of 
price leadership that it could be fairly concluded that rayon producers 
neither tacitly nor explicitly avoid price competition? Unless both of 
these questions could be answered in the affirmative there would be some 
doubts as to the amount of price competition that has characterized the 
domestic rayon industry. Even under the more restrictive interpretation 
of the term "closely related products" the industry would be considered 
workably competitive in periods when the rate of sales is extremely high 
or low relative to installed capacity. But in instances where all firms 
have operated somewhere between 80 per cent and 95 per cent of in- 
stalled capacity and have passively adhered to the price leader's price, 
the industry may not qualify as workably competitive because of the 
second provision of Stigler's definition. 11 

Corwin Edwards has also defined workable competition in terms of 
the structural characteristics in a particular market. They are as follows: 12 

1. There must be an appreciable number of sources of supply and an ap- 
preciable number of potential customers for substantially the same product or 
service. Suppliers and customers do not need to be so numerous that each trader 
is entirely without individual influence, but their number must be great enough 
that persons on the other side of the market may readily turn away from any 
particular trader and may find a variety of other alternatives. 

2. No trader must be so powerful as to be able to coerce his rivals, nor so 
large that the remaining traders lack the capacity to take over at least a sub- 
stantial portion of his trade. 

3. Traders must be responsive to incentives of profit and loss: that is, they 
must not be so large, so diversified, so devoted to political rather than com- 
mercial purposes, so subsidized, or otherwise so unconcerned with results in a 
particular market, that their policies are not affected by ordinary commercial 
incentives arising out of that market. 

4. Matters of commercial policy must be decided by each trader separately 
without agreement with his rivals. 

5. New traders must have opportunity to enter the market without handicap 
other than that which is automatically created by the fact that others are al- 
ready well established there. 

6. Access by traders on one side of the market to those on the other side 
of the market must be unimpaired except by obstacles not deliberately intro- 
duced, such as distance or ignorance of the available alternatives. 

7. There must be no substantial preferential status within the market for any 
important trader or group of traders on the basis of law, politics, or commer- 
cial alliances. 

The structural characteristics of the rayon market since 1932 would 
seem to fulfill the requirements of workable competition by this inter- 

11 This would depend upon whether Professor Stigler considers the temporary 
acceptance of a price leader's price as a "tacit avoidance of price competition for 
fear of retaliation of close rivals." See his footnote, op. cit., p. 3. 

12 Corwin Edwards, Maintaining Competition (New York, 1949), pp. 9-10. 


pretation even under the more rigid definition of an industry. Professor 
Edwards states that there must be an appreciable number of sources of 
supply, but neither suppliers nor customers need to be so numerous that 
each trader is entirely without individual influence. There must, how- 
ever, be no agreement among rivals with respect to commercial policy. 
It would appear, therefore, the oligopolistic markets are ruled out under 
the above set of conditions only when individual firms act in concert to 
an extent that implies either an agreement or the possession of coercive 
power by a large producer. The off-list selling by individual rayon pro- 
ducers in periods marked by a noticeable decline in demand, the dif- 
ferential premiums for quick delivery when yarn shortages appear, and 
the quoted price differentials among producers since the War would 
probably constitute sufficient evidence that rayon price is not a result of 
a collusive agreement. Although American Viscose has generally led the 
industry in effecting list price changes, on several occasions the admitted 
purpose of the price revision was to bring list price more in line with the 
de facto price. Also, several other producers have occasionally taken 
the initiative in revising price schedules. 

These features of rayon price behavior would seem to be compatible 
with the structural characteristics required of a particular competitive 
market as set forth above. There is evidence that each firm considers the 
effect his actions will have upon his rival's behavior, but there is a suf- 
ficient amount of independent pricing over the period of a business cycle 
to give buyers a variety of options among sellers. When this type of price 
behavior appears in a market, fourteen producers probably constitute an 
"appreciable number of sources of supply." It is not likely that Professor 
Edwards would have considered the industry to be workably competitive 
during the 'twenties. From 1920 to 1930 the American Viscose Corpora- 
tion was sufficiently large relative to its rivals to exert a substantial 
amount of coercive power. Also, and this logically follows, the remain- 
ing producers did not collectively control enough productive capacity 
to be able to take over a substantial portion of the trade that usually went 
to the American Viscose Corporation. 

Although it is possible to classify the domestic rayon industry among 
the workably competitive industries by an application of any one of the 
above definitions, it should be reiterated that results obtained from the 
application of each involves a value judgment. Hence, it is unlikely that 
any prescribed set of market characteristics would either obtain the ap- 
proval of all economists or could uniquely determine the workable 
competitiveness of any given industry. Those who reject any set of condi- 
tions that falls noticeably short of the classicists' ideal of perfect com- 
petition would find few workably competitive markets. Those who place 
much reliance upon the dynamics of creative destruction would find 
many more. In the absence of known consequences of an active pursuit 
of either of the foregoing implied policies toward industrial markets, it 


is perhaps just as well that policy making has never been left to either 
of these two groups to the complete exclusion of the other. Even within 
each group that harbors no ideological conflict it is most unlikely that all 
would ascribe to any single set of market conditions universal applica- 
bility. In any case, the workable competitiveness of a particular industry 
is open to debate only after the structural characteristics of its market 
and the dynamic forces that have shaped them have been appraised. 


A relatively small number of large producers (hence, fairly high con- 
centration ratios), short-run price stability, and the acceptance of a price 
leader appear to be the most obvious differences between the rayon 
market and the competitive market of the classicists. In an industrialized 
society in which perfect competition cannot be expected, are those dif- 
ferences sufficiently inimical to the public welfare to conclude that com- 
petition in the domestic rayon industry is "unworkable"? An answer to 
this question involves an appraisal of these differences against the back- 
ground of the dynamic aspects of the rayon industry and an examination 
of the changes that an enlightened public policy would envisage in order 
to eliminate them. 

Although concentration ratios have been high in the rayon industry, 
rapidly declining prices and profit rates, the gradual decline in the rela- 
tive position of the largest and four largest producers, and the increase in 
the number of rayon producing firms over the past twenty-eight years 
are consistent with that which would be expected under conditions of 
competition. In 1920 the American Viscose Company was a highly prof- 
itable (156.83 per cent return on total investment!), patent-protected 
monopoly which sold 150 denier rayon for $4.50 per pound. By 1940 
there were fifteen rayon yarn producers, the average annual rate of re- 
turn for the industry was about 7 per cent, and the price of rayon yarn 
had declined to $0.53 per pound. Over the twenty-year period the quality 
of rayon was greatly improved and domestic output increased from 10.1 
million pounds to 380.1 million pounds per year. Currently, American 
Viscose accounts for only 33 per cent of total domestic output and the 
four largest firms, for about 75 per cent. Two new firms have recently 
entered the industry. Hence, with respect to the number of firms, the 
degree of concentration of productive capacity, and the secular behavior 
of prices, profits and output, competitive forces seem to be accomplishing 
automatically what would presumably be the objectives of an enlightened 
public policy. Profits and prices are declining while output is increasing; 
the number of firms is increasing and the degree of concentration is 

Both the short-run stability and the price leadership which characterize 
rayon price behavior stem from the oligopolistic features of the rayon 
market. It has been previously pointed out, however, that there have 


been a number of significant departures from the general price leader- 
ship pattern. Several downward price revisions admittedly have merely 
taken de jure recognition of de facto prices. Also, although rayon prices 
are more stable than natural fiber prices, there is much interdependence 
among all textile prices. Any appreciable change in silk or cotton prices 
has immediate effects in the rayon market. Since natural fiber prices are 
usually accepted as being highly competitive, it should follow that 
closely related commodity prices geared to them take on some of their 
competitive aspects. In any case, the only obvious way to eliminate prac- 
tices born of oligopolistic rationality is to eliminate the foundations of 
the oligopoly market, i.e., increase the number of independent producers 
in the industry. Hence, the crucial question becomes whether or not 
public policy could be directed toward reducing the market control of 
the larger producers more effectively than it is already being accom- 
plished by competitive market forces. A final decision as to whether the 
rayon industry is workably competitive or not is contingent upon how 
this question is answered. 

Economies of scale in rayon production have been referred to else- 
were in this paper. That the largest plants currently producing rayon 
are the most efficient plants can be established beyond any reasonable 
doubt. It can be fairly safely concluded, therefore, that nothing could 
be gained at present by placing a limitation upon the size of rayon 
plants. Short-run price flexibility, if obtained by this method, is hardly 
worth the social costs involved. But the twenty-eight rayon plants are 
controlled by fourteen firms. Could anything be gained by limiting each 
firm to one plant? An answer to this question hangs largely upon the 
interpretation of all the available evidence. Estimates made by construc- 
tion engineers and the financial statements of individual firms indicate 
that economies of scale are obtained by increasing the size of the firm. 
Nor is there any evidence that any single firm in the rayon industry has 
expanded beyond the point where further economies are impossible. 
Therefore, to farm out all plants to an independent producer would, 
assuming that buyers could be found, entail some duplication in overhead 
(research, physical and chemical testing laboratories, management, sales 
offices, etc.) and thereby raise costs. On the other hand, it is quite likely 
that an increase in the number of rayon producers from fourteen to 
twenty-eight (again assuming that buyers for all plants could be found) 
would eliminate some short-run price inflexibility and make price-leader- 
ship more difficult. However, strong competition from other yarns al- 
ready limits the amount of control the price leader may exercise over 
rayon price. Also, since rayon producers have maintained a high level of 
output in the short run because of cost considerations, they have ren- 
dered inadmissible a number of arguments against short-run price inflexi- 
bility; there is no resultant unemployment to attribute to the "adminis- 
tered" rayon price. Hence, at the present stage of development of the 


domestic rayon industry, it would probably be unwise to place legal 
limitations upon the size of the firm, particularly since the number of 
firms is still increasing. The expected gains to be derived from such a 
policy do not clearly offset the expected losses. 


In concluding that the rayon industry is workably competitive, the 
author lays no claim to having found a more workable definition of 
workable competition. Indeed, he would be among the first to admit that 
others might easily have arrived at totally different conclusions. It might, 
however, not be amiss to state, by way of summary, the guiding princi- 
ples employed in concluding that the domestic rayon industry falls 
within the workably competitive segment of the American economy. 

One of the obvious shortcomings of those definitions of workable com- 
petition which set forth a necessary set of conditions is that they neglect 
the dynamic forces that shape an industry's development. It is difficult 
to visualize a set of static conditions that would not include some men- 
tion of the number of firms. Yet, no one would expect to find "a fairly 
large number" of firms in an industry one or two years after a single 
firm's patent rights to a production process had expired. Should such an 
industry be judged a priori unworkably competitive because it does not 
meet with a stated number of conditions? If changes in the structural 
characteristics of an industry over time are to be admitted to the concept, 
a stated set of conditions is not only inapplicable to all industries at once 
but also loses its applicability to the same industry at different stages of 
development. Further, one might well ask whether it is ever worth- 
while to conclude that a particular industry is not workably competitive 
by a static definition if there are no visible means whereby public policy 
might be directed toward making the industry more competitive. To ap- 
ply rigorously a stated set of conditions might put us in the extremely 
embarrassing position of having caught a host of criminals before we 
have devised an appropriate penal or reform system. 

A possible alternative approach to the concept of workable competi- 
tion may be one which shifts the emphasis from a set of specific struc- 
tural characteristics to an appraisal of a particular industry's over-all 
performance against the background of possible remedial action. Defini- 
tions of workable competition shaped along these lines might accept 
as a first approximation some such principle as the following: An in- 
dustry may be judged to be workably competitive when, after the struc- 
tural characteristics of its market and the dynamic forces that shaped 
them have been thoroughly examined, there is no clearly indicated 
change that can be effected through public policy measures that would 
result in greater social gains than social losses. Tautological through this 
type of definition might be, it at least avoids the pitfall of listing specific 
market conditions that can have very limited general applicability. Also, 


it would ascribe paramount importance to that which should be upper- 
most in the minds of those who formulate public policy — the possibility 
of prescribing appropriate remedial action. For, unless the concept of 
workable competition is to be an instrument of public policy, there is 
little reason for differentiating between workable and pure competition. 
But to frame definitions for public policy purposes without taking cog- 
nizance of the different structural features among industries and within 
the same industry at specific stages of development, and without recog- 
nizing at the outset the political and economic limitations placed upon 
policy-making authorities, would be to ignore the primary purpose of 
such definitions, i.e., to indicate wherein an industry does not operate in 
the public's interest and what appropriate remedial action is possible. It 
seems hardly necessary to point out, however, that definitions of work- 
able competition which follow the above suggested pattern, like all 
others, will not be divorced from value judgments. 

The Decline of Monopoly in the 
Metal Container Industry* 


The Clayton Antitrust Act is now over forty years old, and econ- 
omists are still uncertain about its role in antitrust policy. Most would 
doubtless agree on the need for some instrument that is more flexible 
than the Sherman Act. That Act cannot be applied until monopoly con- 
trol is well established, and one purpose of the Clayton Act has always 
been to prevent monopoly by arresting it in an incipient stage. Its suc- 
cess in this respect has been mixed; many persons have pointed out of 
late that a policy of prohibiting competitive practices that could even- 
tually result in monopoly may be leading us toward per se rules of busi- 
ness conduct that are impossibly confining. But the Clayton Act has also 
been used against business practices in markets that already have a mo- 
nopolistic structure. Sometimes this has been done merely to provide a 
bill of particulars in cases tried chiefly under the Sherman Act; but some- 
times it has been done in an attempt to weaken monopoly through an 
attack on its manifestations. Doubts have also centered on the wisdom of 
this latter policy, on the ground that it does very little good to prohibit 
a market practice if the underlying monopoly power which makes it op- 
pressive remains untouched. 

The American Can case of 1950 illustrates in almost unexpected cir- 
cumstances the force that the indirect and peripheral approach of the 
Clayton Act may have to make competition more effective. This was a 
case in which a specific market practice was the key to market power. 
The practice had nothing to do with the creation of market power, since 
it supervened upon a monopolistic structure that was already there; but 
it played a decisive part in the maintenance of market control after the 
foundations of monopoly had been undermined. 

A cursory examination of the structure of the metal container industry 
just before the antitrust suit began in 1948 might well have made any- 

* The American Economic Review, Vol. XLV (Proceedings of the American 
Economic Association, 1955), pp. 499-508. Reprinted by courtesy of the publisher and 
the author. 

t Vanderbilt University. 



one skeptical of the success of a policy which did not involve funda- 
mental structural reorganization. The industry was highly concentrated, 
and still is: the two leading firms, American Can Company and Conti- 
nental Can Company, together accounted for nearly 80 per cent of total 
sales, and American alone had over 45 per cent. 1 There were in addition 
a number of small sellers — no one of them more than one-eighth as large 
as American — which occupied sheltered and specialized positions or else 
appeared to subsist on a margin of tolerance. The major companies manu- 
facture a product-mix of containers for different purposes, but differenti- 
ation within each product line was and is insignificant. Cross-elasticities 
of demand between metal cans and substitute containers are very small. 
The prices of the two leaders, as well as their other market policies, have 
long been substantially identical. Prices for given products in given sales 
regions were set by one of the large firms, the other following; a policy 
of universal freight equalization eliminated the uncertainties of geograph- 
ical differentiation; and prices exhibited the characteristic rigidity that 
we have come to associate with tacit oligopolistic collusion. 

Tacit collusion in circumstances like these might be expected to re- 
sult in the maximization of joint profits. However, the maximization of 
joint profits is always subject to limitations in any oligopoly structure, 
and in some instances the limitations are so severe that the principle 
loses its explanatory value. Economists have long preferred an eclectic 
approach to this problem. No universal solvent has yet been found that 
will exempt us from a detailed examination of the structure of particular 
oligopolistic markets and their history, as a prerequisite to explaining 
their behavior. The limitations upon monopoly power are often hidden 
from immediate view. A detailed analysis of the metal container industry 
reveals a number of them. 


There is time for only a brief summary here. The first thing to be 
noted is that the limited-duopoly structure in the can industry is a rela- 
tively recent development. Twenty years ago any industrial economist 
would have classed the industry with those dominated by a single firm. 
American Can was organized originally as a trust controlling virtually the 
entire capacity of the industry. It suffered the fate of many of the trusts 
organized at the turn of the century: the umbrella it held over the in- 
dustry encouraged the growth of competition, and within a dozen years 
its share of the market had fallen to 50 per cent. It was about this time, 
in 1916, that American was first charged with monopoly under the 
Sherman Act. Professor Hession in his able study of the food container 
industry (Competition in the Metal Food Container Industry, 1916-1946 

1 United States of America v. American Can Company , Civil Action No. 
26345-H. In the United States District Court for the Northern District of California, 
Southern Division. Exhibits UU and 3124 (Stipulation on Statistics, March 31, 1949). 


[Brooklyn, 1948] ) calls American Can "the trust that wasn't busted." The 
court refrained from dissolving it principally because of the rapid decline 
in its market share, and the judge expressed the hope that the growth of 
rivals would soon restore competition in the industry. 

Competition was not quickly restored. American possessed nothing 
approaching pure monopoly after 1916, but it remained the dominant 
firm. Its market share ceased to fall, and its capacity remained for a 
time enormously greater than its rivals'. American's prices determined 
the prices for the whole industry. It maintained its lead by rapid develop- 
ment and expansion from within. Nevertheless, its power to control the 
industry was considerably weaker in 1948 than it was in 1916. 

This slow metamorphosis was due partly to developments within the 
industry, partly to the impact of the antitrust laws. The structure of the 
industry altered gradually as Continental Can began to challenge Ameri- 
can's leadership. In 1916, Continental was a small cloud on American's 
horizon, hardly distinguishable from a number of other small rivals. 
Throughout the following decades, Continental grew rapidly with a bold 
program of acquisition and merger as well as by rapid expansion from 
within. By 1939, it was half as large as American; by 1950 it was three- 
fourths as large. Its growth eventually put it in a position that rivalled 
American's. Prior to the middle thirties, American was the only metal 
container manufacturer with an important research program and was 
practically solely responsible for the progress and development that had 
occurred up to that time. Moreover, it was the only manufacturer up to 
then that was fully able to realize certain other advantages of scale. 
These are not production economies, but are achieved through the inter- 
locking of separate plants so that the extreme uncertainties of demand for 
food containers in any one area are diffused along a chain of production 
sites. The ability of American to offer certainty of supply, together with 
its facilities for research and customer service, made it the only accept- 
able source of supply to many buyers of cans, including most of the 
large and stable ones, until a second large seller appeared. 

By the end of the thirties, American was at last confronted with a 
powerful, efficient, and progressive rival in practically every geographi- 
cal region and every product, including many in which American had 
earlier enjoyed a monopoly. It took time for Continental to break into 
the geographical strongholds of American and to build up an effective 
and experienced research staff, but in the end the supremacy of Ameri- 
can was undermined. 

Conditions were more stringent for the smaller firms. No other was 
able to achieve anything like the growth of Continental; in fact, several 
of those which had begun to grow to substantial size were acquired by 
Continental. A few producers of food cans managed to build up inte- 
grated regional organizations, securing at least part of the advantages of 
size, and compensated for their inferior research and service facilities 


by charging somewhat lower prices. Two firms even succeeded in enter- 
ing the industry and establishing themselves in the food can market. All 
of these smaller producers were potential bases for expanded compe- 
tition; that this potential was not fully realized was due to certain market 
practices of the larger firms which held the smaller ones in check. 

A significant change was forced by the tightening of the law against 
price discrimination after 1936. The Robinson-Patman Act affected two 
markets: tin plate and tin cans. Tin plate accounts for 60 per cent of 
metal container costs. In the days of the Steel Trust and the Can Trust the 
tin plate market was almost a classic bilateral monopoly. The monopolies 
on both sides lost their near-exclusive control, but the "leading bargain" 
between U.S. Steel and American Can continued to determine the price 
of tin plate. In this market it appeared that monopoly power was offset 
by monopoly power. The benefits of this countervailing power, however, 
were mixed. Improvement and development in the tin plate industry, 
for instance, have been rapid and considerable, and there is no doubt that 
the large buyers, especially American, have pressed the steel producers 
closer to the maximum attainable rate of progress — both in product im- 
provement and cost reduction — than would have been the case if buyers 
had been small and impotent. On the other hand, the price benefits of 
countervailing power filtered down to consumers very slowly; it requires 
a substantial degree of competition in subsequent markets before counter- 
vailing power in a prior market amounts to anything more than a division 
of spoils. One result of American's dominant position as a buyer of tin 
plate in the early years was that the tin plate producers favored it with 
secret rebates and systematic price discrimination. After a second buyer 
of substantial size had entered the market a new dimension of strategy 
emerged, and price reductions negotiated between individual buyers and 
sellers became more difficult to confine to the largest buyers. Neverthe- 
less, systematic discrimination continued in various forms up to the pas- 
sage of the Robinson-Patman Act, at which time American actually 
capitalized its claims to discriminatory treatment and liquidated them by 
accepting substantial lump-sum settlements from United States Steel and 
other suppliers. After 1937, countervailing power was exercised in behalf 
of all buyers of tin plate. This benefited the ultimate buyers of cans only 
insofar as there were limits to the exercise of monopoly power in the can 

The buyers of cans were themselves not entirely helpless, though 
none had anything like the bargaining power that American had in the 
tin plate market. Even in the days when there was only one large can 
manufacturer, the big canning companies had the option of manufactur- 
ing for themselves. Backward integration is not extremely difficult for 
a large user, and it would have taken only a slight widening of the profit 
margin on cans to make it economical. Potential entry from the buying 
side limited monopoly power on the selling side. At first the result was 


systematic, secret discrimination in favor of the buyers who could offer 
this threat. After a second seller grew to large size, the bargaining posi- 
tion of the large buyers was strengthened, since they could play one off 
against the other. The Robinson-Patman Act obliged the major can com- 
panies to abandon secret discrimination. It was replaced, not by uniform 
prices to all buyers, but by an open schedule of volume discounts, which 
eventually came to be based on total purchases of all kinds of containers 
from the supplier. The discounts were greater for large-volume purchas- 
ers than any concession that the smaller can companies could easily offer, 
and there is no doubt that these discounts worked to their disadvantage. 
On the other hand, the continuous alteration and reshaping of the dis- 
count structure did afford a means of limited price competition between 
the two leaders; and the large buyers had at least the potential power to 
enforce this competition between them. 


Though the monopoly power once possessed by the dominant firm 
had been markedly eroded by 1950, the effect of the underlying struc- 
tural changes on competition remained more potential than actual. The 
reason was that the leading firm followed several policies which con- 
centrated its remaining market power in the most effective way, and its 
large rival did likewise. There were three important instruments of com- 
mercial dominance. The first was the volume discounts just mentioned. 
The second was the practice of selling under long-term requirements 
contracts. It is fair to say that neither of these could have had a power- 
ful enough effect to justify action under the antitrust laws, though they 
did tend to reinforce each other. The contracts were written for spe- 
cific containers for use at specific plants. Smaller can companies occa- 
sionally succeeded in becoming secondary suppliers of large buyers, but 
it was the usual practice for a buyer to concentrate all his purchases on 
a single source of supply. There were some buyers whose purchases con- 
siderably exceeded the volume necessary to get the highest discount that 
was offered, but most of these also had long-term requirements contracts 
with a single supplier or, less frequently, split their purchases between 
American and Continental. The amount of business that the smaller can 
companies were eligible to compete for at any given time was thus re- 
stricted. Volume discounts were partly responsible, but the superiority 
of the two large sellers in research and customer service may have been 
equally important in determining the result. 

The third instrument of control was the practice of tying the lease 
of can closing machinery to the sale of cans. (This was not done 
through an explicit tying clause, but by arranging the expiration dates 
of separate contracts so that no canner would be able to retain Ameri- 
can's machinery to close competitors' cans. There were minor excep- 
tions.) This requires a word of explanation. If the manufacture of clos- 


ing machinery had been perfectly competitive, the larger can companies 
would have gained no net advantage by this practice. But American's 
can machinery has long been recognized as the best and most complete 
in the industry, and its progressiveness in this field is beyond dispute. 
Continental began late, but was overtaking American rapidly in the late 
thirties. Both large firms followed the same policy. Rentals on closing 
machinery, which also covered servicing by the manufacturer, were set 
below actual cost, to induce buyers to lease machines. Most can buyers 
were allowed to retain the machinery only as long as they purchased the 
cans to be closed on it from the same firm. The moderate advantage 
which the major sellers possessed in closing machinery was thus extended 
forward and magnified in the can market. The relation between supplier 
and customer became harder to break. (Closing machinery is generally 
unimportant for nonfood cans, and both major companies faced much 
more vigorous competition from independents in this field.) In markets 
for food cans, the smaller manufacturers were forced to imitate the 
leasing practices of the majors; but since they had to buy somewhat in- 
ferior closing machines on the open market and lease them at below-cost 
rentals, they reaped no comparable advantage. (One or two of the smaller 
companies also entered the manufacture of closing machinery, without 
much success until recently.) A side effect of the practice was that no 
canner would buy machinery in preference to renting it, and independent 
machinery manufacturers faced an uneconomically thin market. 


The government might have attempted a major Sherman Act case, 
with wholesale dissolution and structural reorganization of the industry 
as the objective. But this would have been risky, since the specific 
charge of monopolization had been tried in an earlier case and the judge 
had refused the suggested remedy of dissolution. Instead, the government 
attacked the practice of tie-in sales directly under Section 3 of the Clay- 
ton Act, retaining a Sherman Act charge to scoop in the requirements 
contracts if possible and to provide for equitable relief. There is some 
evidence that the Antitrust Division was not entirely satisfied with the 
form that its case eventually took in consequence, but it won the case 
nevertheless. The court decided that American had violated both the 
Clayton Act and the Sherman Act. 2 In addition to a number of minor 
matters, which need not concern us here, the 1950 decree altered market 
practice in the metal container industry in three ways. (Continental ac- 
cepted the same judgment in a consent decree.) 

1. The major firms were prohibited from offering any annual cumula- 
tive volume discounts. 

2 87 F. Supp. 18 (1949). A simultaneous case against Continental was not brought 
to trial, Continental having agreed in advance to accept any judgment entered 
against American, short of divestiture. 


2. Requirements contracts were limited to one year. American Can 
had presented convincing evidence that requirements contracts are nec- 
essary to protect food canners, in view of the great uncertainty of crop 
yields and timing, and that canners preferred them to fixed-quantity 
contracts; hence the court did not prohibit them altogether. The court 
also ordered that separate contracts be written for individual plants when 
the customer had more than one cannery. The buyer could still place all 
his contracts with a single seller if he chose, but smaller can manufacturers 
would find it easier to bid at least for the business of a single plant of a 
multiplant buyer. 

3. The tie between machine leasing and the sale of cans was broken. 
The major companies were permanently enjoined from conditioning the 
lease of machines on the sale of cans, by any subterfuge. In addition, 
they were ordered to sell their existing closing machines at bargain 
prices to anyone who wanted to buy them, giving priority to existing 
lessees. This order applied to all machines to be built in the future as 
well, for a period of ten years. The can companies were required to 
make all technological information and know-how available to buyers, 
to set up schools to train service men employed by the canneries, and 
to license closing machinery patents without royalty. The court recog- 
nized that there were certain advantages to integration between machin- 
ery manufacture and can manufacture, since can manufacture and can 
closing involve much the same technology, and so it did not order divesti- 
ture. The can companies were required to lease to everyone any ma- 
chine that they were then leasing to anyone. Moreover, rentals had to be 
fully compensatory, including a fair profit, after the end of 1953. (Conti- 
nental has asked for a postponement. Fully compensatory rentals on its 
diminished stock of equipment would be prohibitive, it claims, since 
overhead has not been proportionally reduced.) No more could below- 
cost rentals be used as an inducement to lease. 

This decree knocked out practically all the remaining props of market 
control. The decree has been in effect now for over three years. Some 
of its consequences have been striking. One instance is the sale of clos- 
ing machines, which has greatly exceeded everyone's expectations. The 
can manufacturers had claimed that customers would be reluctant to buy 
them — that customers would prefer to lease instead and pay the supplier 
the appropriate fee for bearing the risk of obsolescence and the task of 
servicing and maintenance. If this reluctance ever existed it was quickly 
overcome. The bargain prices the court set on the machines existing in 
1950 made it uneconomical not to buy them, and in addition American 
and Continental, in wholehearted compliance with the letter and the 
spirit of the decree, have made vigorous efforts to sell both old and 
newly manufactured machines. By the middle of 1954 both American and 


Continental had sold over 75 per cent of the closing machines they were 
leasing in 1950. While the lease market may revive in the future, can buy- 
ers are no longer heavily dependent on their can suppliers for closing 
machinery, and there is no way now for the machinery supplier to apply 
commercial leverage to the can market. After the transition period is 
over, the independent manufacturers of machinery will be able to market 
closing machines directly to the canning industry, and they have already 
responded to this broadening of their opportunities. 

A second result has been an extensive breakdown of exclusive supplier- 
customer relations. While fruit and vegetable canneries have largely 
continued their requirements contracts, the large packers operating sev- 
eral canneries have begun to allocate them to different suppliers. Other 
canners have begun to split requirements within the plant, closing cans 
from several suppliers indiscriminately on the machines of several manu- 
facturers. There has been an enormous growth in open-order purchasing 
at the expense of contract purchasing. Small firms are now better able 
to detach fragments of business which used to be held firmly by the 
large can suppliers. They can enter large-volume markets either by 
carrying off supply contracts for individual canneries in a chain or by 
becoming secondary sources of supply within the plant. The vertical 
partitions in the market which formerly made the ties between particular 
buyers and sellers very strong have been demolished, and competitive 
forces, wherever they originate, can sweep across it largely unimpeded. 
The market position of smaller manufacturers has been greatly strength- 

It is interesting to note that these changes in the can market have 
been accompanied by some instability in the tin plate market. At least 
part of this instability has been transmitted backward from the can 
market. Prices have become more flexible; open-order purchasing has 
grown in volume; the leadership of the leading bargain between U.S. 
Steel and American Can has weakened further. The tin plate suppliers 
still feel the weight of the large buyers' bargaining power, but this is 
more likely now to work to the benefit of ultimate consumers of cans. 

It is possible of course to expect too much of the decree. The gross 
structure of the industry has not changed very much and probably will 
not change much in the immediate future. Doubtless the effects of a 
decree of dissolution, if American had been convicted of illegal acquisi- 
tion and maintenance of monopoly, would have been more spectacular. 
Several firms could have been fashioned out of either of the leaders 
without an appreciable loss of efficiency, and the market would in time 
have enforced a high degree of competition among the fragments. How- 
ever, such a monopoly charge could not have been sustained. We must 
decide whether we have any better ground now than the court did in 
1916 for expecting workable competition in the future. 

American and Continental together would still be able to dominate 


the industry in the short run if they maintained effective collusion in 
every dimension of the market. But instead the two large firms are in a 
state of intense rivalry. The challenger is staging a vigorous drive on 
the markets of the leader. A number of large accounts previously held 
by American have now been split. Technological rivalry is unrestrained. 
While open price warfare has been avoided, except in a few local in- 
stances, there is protracted maneuvering in an atmosphere of great un- 
certainty when the time for quoting new prices and renewing contracts 
comes round every year. American's price leadership is now merely 
barometric, and it cannot count on being followed by Continental or by 
any other seller. 

The smaller firms, aided by a strong growth trend in the industry, 
have been establishing themselves on a more secure footing and have 
also been edging into the markets of both the leaders. In several in- 
stances recently the price structure of the majors has had to be adapted 
to the independent competitive tactics of a smaller firm aggressively 
reaching for a larger share. Although the oligopoly structure remains, 
enough has been said to show that the competitive pattern is in a state 
of flux. 

No outsiders have entered can manufacture in the last three years, 
but the threat of entry from the buying side has been intensified. Large 
buyers can no longer be pacified with volume discounts. During the 
past year the very largest consumer of packers cans began to manu- 
facture part of its own requirements, and another large buyer, located in 
one of American's few remaining monopoly territories, has announced 
plans to manufacture all its own containers. The effect of these events 
is incalculable. Both of the large sellers are constrained as never before 
by the threat of backward integration, and the restraints exercised on 
profit margins in cans work to the benefit of all buyers. 

American itself has no monopoly weapons left in its hands. Any firm 
which relies on size and the momentum of past achievements alone to 
protect it against competition is likely to find its position deteriorating 
with alarming speed. What American can do is to take advantage of the 
moderate superiorities it evidently has in efficiency and research. It can 
maintain its leading position by reducing costs and continuing its rapid 
rate of product development, passing along the benefits to consumers. 
Its large rival will push it, or perhaps lead it, in this respect. Its smaller 
rivals will offer a stronger competitive challenge as time goes on. And 
its large customers will never be more than a few years behind in poten- 
tial efficiency if they should decide to produce their own supplies. A 
forecast of workable competition appears to be justified. Thus it seems 
that the Antitrust Division, in choosing to make a limited attack on 
market practices — however much it may later have doubted its own 
wisdom — made a good decision after all. 


The Tobacco Case of 1946 


"If the law supposes that," said Mr. Bumble, ". . . the law 
is a ass, a idiot." 

— Charles Dickens, Oliver Twist 

The ways of the law are devious and unpredictable. Since the law is 
(in Tennyson's words) a "wilderness of single instances," it is dangerous 
to assume that future court decisions will always be fully consistent with 
those of even the recent past. Yet, being neither prophet nor crystal- 
gazer, I must credit the courts with a logical consistency which I know 
they do not possess. I can therefore discuss even the probable economic 
consequences of the Tobacco case only upon the basis of two dubious 
assumptions: (1) that the courts really said what I believe them to have 
said in the Tobacco decision; and (2) that they will carry to their logical 
conclusion the legal implications of that decision. Having satisfied my 
conscience by this urgent prefatory caveat, I can more honestly turn to 
the difficult task at hand. 


It is the task of the law of monopoly to distinguish between business 
practices which are in the public interest and those which are not. In 
carrying out this difficult problem of evaluation, the courts have had to 
devise and apply tests capable of differentiating between approved and 
disapproved practices. As elsewhere in the law, the law of monopoly has 
reflected the perennial conflict between certainty and change. Two tests 
of monopoly have become traditional: (1) On the question of conspiracy, 
does the evidence show that competitors actually agreed? (2) On the 
question of monopolization, was there overt predatory action to exclude 
competition? These two tests had the advantages of certainty — they 
could be applied with sufficient consistency to assure equality of treat- 
ment before the law; and they were sufficiently concrete to indicate the 
practices which must be avoided to escape condemnation under the law. 

* The American Economic Review, Vol. XXXIX (Proceedings of the American 
Economic Association, 1949), pp. 284-96. Reprinted by courtesy of the publisher and 
the author. 

I am indebted to my colleague, George W. Stocking, for his helpful but some- 
times dissenting criticisms of the original manuscript of this paper. 

t Vanderbilt University. 



Unfortunately, however, these tests have become increasingly inadequate 
as the structure and practices of American industry have taken new and 
more subtle forms. Thus, the need for change — for adapting the law to 
a new industrial environment — has become more and more apparent. 

In the Tobacco case, 1 at the necessary cost of new uncertainties, the 
courts finally met this need for change in two ways. First, the Court 
of Appeals brought wholly tacit, nonaggressive oligopoly fully within 
the reach of the conspiracy provisions of the Sherman Act. Prior law 
already had made clear that — in the absence of "formal agreement" — 
an unlawful conspiracy can be inferred from "concert of action," "unity 
of purpose," or "a common design." Furthermore, in the Tobacco case, 
there was plentiful and undisputed evidence that the three defendant 
dominant firms had behaved identically with regard to prices, terms of 
sale, and general business practices. Nevertheless, the case was probably 
unique in that there was not a whit of evidence that a common plan had 
even been contemplated or proposed. 2 The government's evidence was 
admittedly wholly circumstantial. The fact of identity of behavior was 
offered as the basis for inferring both the existence and the elements of 
the alleged common plan and the defendants' knowledge of that plan. 
Each was alleged to have acted similarly with the knowledge that the 
others would so act, to their mutual self-interest. Thus, the Tobacco 
case brought the basic assumption of modern oligopoly theory squarely 
before the courts. In finding in the facts a reasonable basis for the 
jury's inference of unlawful conspiracy, the Court of Appeals accepted 
the practical implications of that assumption; namely, that a few domi- 
nant firms will, perhaps independently and purely as a matter of self- 
interest, evolve nonaggressive patterns of behavior. Thus, attention was 
shifted from form to probable results. Upon final appeal, the Supreme 
Court refused to review this part of the lower court's findings. 

Second, in the Tobacco case, the existence of power to exclude com- 
petition, not the abuse of that power, became the new test of illegal 
monopolization. Thus, the recent Aluminum doctrine 3 was approved and 
was extended to conspiratorial oligopoly. Accepting without review the 
judgments below that a conspiracy had been established, the Supreme 
Court held that neither the exertion of power to exclude nor the actual 
exclusion of competitors is necessary to the crime of monopolization. 

^American Tobacco Co., et al., v. U.S., 147 F. 2d 93 (1944); 328 U.S. 781 (1946). 

2 In this regard, it therefore differed from the Interstate Circuit case (306 U.S. 
208, 1939), upon which the government relied. There it was held (at 226) that the 
finding of an agreement among distributors, while supported by the evidence, was 
not necessary: "It was enough that, knowing the concerted action was contemplated 
and invited, the distributors . . . participated in it." In that case, however, unanimity 
of action would have served only as evidence of adherence to a common plan, the 
proposal of which was supported by direct evidence. 

3 U.S. v. Aluminum Co., 148 F. 2d. 416 (1945), the Circuit Court serving as the 
final court of appeal. 


Existence of the power and intent to exclude will suffice. Since the Court 
was willing to infer intent from the concerted action of the conspirators, 
the power to exclude — as shown by the degree of market control of the 
combination — was made the crucial issue. Thus, attention was shifted 
from monopolization as an action to monopoly as a condition. The Su- 
preme Court explicitly limited the precedential significance of this deci- 
sion to cases in which conspiracy is an essential ingredient. Nevertheless, 
so broad was the lower court's application of the law of conspiracy to 
the facts that, if it were generally followed, the behavior of few oligopo- 
lies could probably escape condemnation as "conspiratorial." 

It should be obvious that the Tobacco decisions of 1946 must be given 
a prominent place in the historical development of the case law of the 
antitrust acts. And, if my interpretation of their legal implications is cor- 
rect, the Tobacco decisions have gone far to close the wide gap between 
the legal and economic concepts of monopoly, which became so apparent 
to economists during the thirties. 4 In accepting detailed similarity of be- 
havior among a few dominant firms as a reasonable basis for inferring il- 
legal conspiracy, the courts have finally brought the law of conspiracy 
into harmony with the economics of oligopoly. Furthermore, the legal 
and economic concepts of monopoly (including conspiratorial group 
monopoly) have been brought closer together. Thus, Rostow is correct 
in finding, on the basis of the Aluminum and Tobacco cases, that "market 
control is now a far more important theme in Sherman Act cases than 
handicaps on an individual's power to do business. The old preoccupation 
of judges with evidence of business tactics they regarded as ruthless, 
predatory and immoral has all but disappeared. . . . We are close to the 
point of regarding as illegal the kind of economic power which the 
economist regards as monopolistic." 5 That the Aluminum and Tobacco 
decisions have revitalized the Sherman Act, especially Section 2, seems 
beyond serious doubt. Furthermore, economists can rejoice that the pre- 
cepts of modern economic analysis have so quickly found their way into 
the case law of monopoly. 

Unlike lawyers of the Antitrust Division, however, economists cannot 
be satisfied simply by the scoreboard of cases won by the government. 
Although some important legal "twilight zones" remain, the widespread 
application of the new tests of monopoly to oligopolistic industries 
should lead to a high proportion of government victories from now on. 
It is clear, however, that without remedial action, these legal victories 
will be of relatively little economic or social consequences. And, if 

4 Cf. E. S. Mason, "Monopoly in Law and Economics," Yale Law Journal, 
Vol. 47 (1937-38), pp. 34-49. 

5 Eugene V. Rostow, "The New Sherman Act: A Positive Instrument of Prog- 
ress," University of Chicago Law Review, Vol. 14 (1946-47), pp. 574-75 et seq. 
For a more cautious appraisal of these two cases, cf. Edward H. Levi, "The Anti- 
trust Laws and Monopoly," ibid., pp. 172-81. 


remedial action is taken, it is by no means certain that the economic con- 
sequences will always be in the public interest. 


With regard to the Tobacco decision, counsel for Reynolds Tobacco 
Company argued that: 

. . . the significance of these convictions extends far beyond the immediate 
consequences to petitioners and the tobacco industry. . . . For, if these convic- 
tions be lawful, the pattern of prosecution is applicable — with the result of al- 
most certain and repeated conviction — to every other executive and corpora- 
tion in a mass production industry ... in which, as a matter of common 
knowledge, economic forces have produced identities or close similarities in 
manufacturing, packaging, pricing, advertising, marketing and even raw mate- 
rial acquisition. 6 

The present writer is willing to accept this appraisal as being at least 
within the realm of possibility. Given the widespread pattern of domi- 
nation-by-a-few in modern American industry, counsel for Liggett and 
Myers was also probably correct in arguing that "the common practices 
of the tobacco industry are in many instances usual features of business 
life today, and in all instances practices which businessmen guided by 
. . . self-interest, acting reasonably and in the absence of agreement, 
might adopt." 7 In view of this fact, the conviction of the major tobacco 
companies suggests at least a presumption in favor of the view that the 
Antitrust Division's ability to find and prosecute monopolies successfully 
is now largely limited only by the extent of its own resources in bringing 
cases to trial. 

Nevertheless, we must recognize that at least two important questions 
are as yet unanswered. First, what types of similarity of behavior among 
oligopolistic competitors will the courts hold to be insufficient to sus- 
tain an inference of illegal conspiracy? In the Tobacco case, the circum- 
stantial evidence supporting such an inference was very strong — con- 
siderably stronger than the evidence which could be marshaled against 
many other oligopolistic industries. However, the extent to which the 
courts are willing to go in finding illegal conspiracy among oligopolists 
will depend at least as much upon their judgment and preconceptions as 
upon the facts of any specific case before them. The conspiracy doctrine 
of the Tobacco case certainly permits them to go about as far as they 
like in this direction. Second, what is the legal status of a single domi- 
nant firm of intermediate size (say, controlling 50-65 per cent of an in- 
dustry), such as was involved in the Steel and Harvester cases? Until 
the courts explicitly apply recent doctrine to a situation of this sort, the 
law will continue to treat loose "conspiracies" more severely than (within 

G In the Supreme Court of the U.S., October Term, 1944, No. 840, Reynolds v. 
U.S., Petition for Writ of Certiorari, p. 14. Italics added. 

7 In the U.S. Circuit Court of Appeals, 6th Circ, No. 9138, Liggett and Myers 
v. U.S., Brief on Behalf of Appellants, p. 234. 


certain limits) such a single dominant firm with an equal or greater de- 
gree of market control. 8 

Despite these remaining "twilight zones," the conclusion is apparent: 
in antitrust action against oligopolistic industries, the prospects of gov- 
ernment victory are now relatively bright. 


It is the writer's belief that, in the Tobacco case, the courts reached 
a conclusion which — in the main — economic analysis would support. 
Nevertheless, the fact remains that the court in effect condemned the 
natural, normal, and intelligent consequences of an oligopolistic market 
structure. The difficult question of appropriate remedial action was 
therefore placed in bold relief. As a criminal prosecution, the Tobacco 
case provided no remedial action. Thus far, therefore, the government's 
victory has been (apart from a quarter-million dollars of fines) almost 
an empty one. 

Given the present structure of the industry, until now left untouched 
by the government, the writer must admit a feeling of some sympathy 
for the arguments of counsel for the tobacco companies. Thus, counsel 
claimed that, even if guilty, they were "entirely without guide as to 
how they may lawfully avoid the creation of evidence of future Sher- 
man Act violations against themselves, unless they cease business al- 
together." For example, must Reynolds refrain from "percentage buy- 
ing" of leaf? And if it does, and its competitors do likewise, "will it or 
they not then be accused of manipulating prices, allocating tobaccos, 
and discriminating against growers through intermittent buying ....?" 

8 Reynolds' counsel posed essentially this same problem in the following argu- 
ment, even though (written prior to the final Aluminum decision) it was based 
upon a false premise. Since it was not denied that "exclusionary action is essential 
to monopolizing by a single person or entity": 

"It should a fortiori be held essential in a case involving a number of separate 
entities severally engaged in competition (albeit imperfect) with each other. Other- 
wise, the whole notion of group monopolizing becomes illusory. Without concerted 
action to exclude others from the field, there is no justification for analogizing the 
position of the several entities to that of a single trading unit seeking sole, or 
practically exclusive, trading privileges. Nor is there any warrant for aggregating 
their power in order to show control over the field. For, until action to exclude is 
taken, the power of none is employed to fence in the field in the interest of the group. 
On the contrary, the power of each is being employed in competition with the 
others and constitutes a disruptive or centrifugal force. And an unexecuted intent to 
exclude someone else at some indefinite time and by some undetermined means does 
not serve to fuse that power and make of it a centripetal force." (In the Supreme 
Court of the U.S., October Term, 1945, Reynolds v. US., Brief for Petitioners, 
pp. 47-48.) 

The extent to which the Aluminum doctrine is applicable to single dominant 
firms in other industries is obscured by the fact that Alcoa's percentage control was 
very large (90 per cent). In the Aluminum opinion, Judge Hand himself expressed 
doubt that a single firm controlling 60-64 per cent of a market was a monopoly; 
he was certain that 33 per cent was not. Significantly International Harvester's market 
position had been 64 per cent. 


Or "must Reynolds desist from charging for its product the price 
charged for a competitive product, and must Reynolds, by prosecution 
or otherwise, attempt to prevent a competitor from selling at Reynolds' 
price?" 9 Or again: 

What are the specific policies and practices we must abandon, modify, or 
adopt in order to conduct our business according to law? . . . Presumably, the 
appellants were convicted of agreement, not of the particular operations alleged 
to constitute agreement. Yet, on the Government's theory, continuation by 
more than one of the appellants of the operations alleged is evidence of a 
further Sherman Act agreement. ... If this is so, how is Liggett & Myers to 
carry on? Must it start all over again with new management, with a new 
system? Is everything the appellants do illegal, or evidence of illegality, if done 
by more than one of them? 10 

Since neither the prosecution nor the courts provided an answer, the 
major tobacco companies have themselves done so; namely, to follow 
essentially the same cigarette price policies since the trial that they fol- 
lowed before. The evidence upon which the Tobacco decision was based 
extended only through July, 1940. Effective July 1, 1940, the three major 
cigarette companies had increased their list prices by identical amounts 
from $6.25 to $6.53 per thousand cigarettes to take account of an in- 
crease in the federal cigarette tax from $3.00 to $3.25 per thousand. As 
a result, their net realized price after discounts to wholesalers (10 and 2 
per cent) and federal tax was virtually unchanged (dropping from $2,512 
to $2,509 per thousand). Although the price history of the industry 
since 1940 has been dominated by price controls, the recent behavior of 
cigarette prices is of considerable interest. On December 27, 1941, Amer- 
ican raised the list price of Lucky Strikes from $6.53 to $7.10. This 
represented the first time since the period of virtual list price identity 
began in 1923 that American had attempted to lead upward in a price 
change although it had twice led downward in 1933. Reynolds and Lig- 
gett and Myers refused to follow, however. Whether their failure to do 
so was the result of their unwillingness to recognize American as the 
price leader is impossible to determine because of two other important 
factors in the picture at that time. First, the three companies had just 
filed an appeal to the unfavorable district-court verdict which had been 
largely based upon identity of price behavior. Second, Price Administra- 
tor Henderson immediately asked American to rescind its increase pend- 
ing investigation of the increased costs by which the increase was al- 
legedly justified. When American refused unless formally ordered to do 
so, the OPA on December 30 froze all cigarette prices at their Decem- 

9 In the Supreme Court of the U.S., October Term, 1944, No. 840, Reynolds v. 
U.S., Petition for Writ of Certiorari, pp. 12-13. Cf. ibid., Liggett and Myers v. U.S., 
pp. 5, 26. 

10 In the Circuit Court of Appeals, 6th Circ., Brief on Behalf of Liggett and 
Myers, op. cit., p. 27. 


ber 26 levels, although it permitted the list prices of the economy brands 
to be raised from $5.05 to $5.15. n 

Although the price ceilings on the economy brands were gradually 
raised, OPA did not permit any increase in the prices of the standard 
brands (except one to compensate for the increase of $0.25 in the federal 
tax on November 2, 1942) until April, 1946. As a consequence, the net 
price to the major companies, after discounts and federal tax, remained 
virtually constant at $2.51 from January, 1937, to April, 1946. During 
this same period, with rapidly rising manufacturing costs, the three com- 
panies' net incomes as a percentage of their net worths (after deducting 
book value of good will) fell from 15-20 per cent to 9-11 per cent. On 
April 26, 1946, OPA raised the price ceilings on standard and economy 
brands to $7.09 and $6.10, respectively. 

On July 1, 1946, existing price controls expired and the new act of 
July 25 exempted tobacco products. On July 30, Liggett and Myers 
announced a price increase of $0.22 per thousand on Chesterfields, but 
rescinded it retroactively a week later when other manufacturers refused 
to follow. 12 Thus, in its first attempt since the dissolution to lead in a 
price change, Liggett and Myers was unsuccessful. That the other com- 
panies did not follow may have reflected their unwillingness to concede 
a position of price leadership to Liggett and Myers. On the other hand, 
fear of adverse public opinion in the turbulent days of mid- 1946 may 
have been a factor. On October 7, however, American raised the price 
of Lucky Strikes from $7.09 to $7.38. Liggett and Myers' Chesterfield 
followed upward by an identical amount within a day or two. Finally, 
on October 11, after most other manufacturers had also followed Amer- 
ican's lead, Reynolds increased the price of Camels to $7.35, three cents 
per thousand below the other major brands. This minute price differential 
was sufficiently unique to attract considerable notice in the press. With 
this minor exception, all standard brands were apparently raised to an 
identical higher list price. 13 

On July 28, 1948, American again led in a price increase from $7.38 
to $7.78. All of the other major companies followed within twenty-four 
hours, the prices of Camel and Philip Morris being raised to $7.75 and 
$7.79, respectively. It is believed that Chesterfields were raised to $7.78. 
Up to the present time, however, the writer has not been able to obtain 

^Business Week, January 3, 1942, p. 8; 7 Federal Register 1322. 

12 Wall Street Journal, July 31, 1946, p. 14; ibid., August 6, 1946, p. 4. 

13 These price data are from the Wall Street Journal (October 8, 1946, p. 14; 
October 9, 1946, p. 4; October 12, 1946, p. 2), which reported that the following 
other brands were all raised to $7.38: Pall Mall and Tareyton (American) ; Old Gold 
(Lorillard); Philip Morris (Philip Morris), increased the same day as Lucky Strike; 
and Rameses (Stephano Bros.). Marvel (Stephano), a leading economy brand, was 
also increased from $6.10 to $6.40. The price changes on Brown & Williamson brands 
(Raleigh, Kool, Wings) were not listed. 


the exact list prices of the other major brands (including Chesterfield), 
but he is certain that they were raised by almost identical amounts. 14 
Insofar as price differences among the standard brands do now exist, 
it is obvious that they are of no practical importance, being too small to 
be reflected in different retail prices. Furthermore, there have been no 
changes in the customary discounts to wholesalers. Hence, even though 
very minor price differences may have resulted from the Tobacco deci- 
sion, they are significant only as a means of attempting to avoid the 
absolute price identity there condemned. In fact, with the minor excep- 
tions noted, the list prices, terms of trade, and timing of price changes 
are probably more nearly identical today — if one considers all of the 
standard brands — than prior to 1940. 15 

This limited evidence supports the view that, if there was illegal 
conspiracy before 1940, there is still illegal conspiracy today. Further- 
more, if the economic power of the three major companies combined 
constituted illegal monopolization before 1940, it does so a fortiori at 
the present time. Between 1939 and 1947, during which domestic ciga- 
rette sales nearly doubled, the three defendant companies expanded their 
share of the domestic cigarette market from 68 to more than 85 per 
cent. Thus, they appear to be fast approaching their position of 1931, 
when they sold 91 per cent of the nation's cigarettes. Meanwhile, the 
economy brands' share of the domestic market has dropped from 14 to 
about 1 per cent in the face of high consumer incomes, high manufactur- 
ing costs, and a narrowing price differential below the standard brands. 
As a consequence, the smaller companies which grew rapidly on econ- 
omy brands during the thirties have lost much ground since 1939, their 
standard brands (if any) failing to share sufficiently in the shift to a 
higher-priced product to compensate for their declining low-priced 
market. Apart from Philip Morris (which has not quite held its own 
despite its purchase of Axton-Fisher in 1944), all companies other than 
the Big Three have suffered moderate to severe losses of market posi- 
tion since 1939. Within the Big Three, Liggett and Myers' share dropped 
slightly while American's jumped from 22.7 to 36.1 per cent, and Reyn- 
olds' from 23.7 to 28.8 per cent. Not since 1931 have one or two com- 

14 These price data are from the Wall Street Journal (July 30, 1948, p. 12; July 31, 
1948, p. 2) and the New York Times (July 29, 1948, p. 23; July 30, 1948, p. 11). 
The prices of the two principal economy brands, Marvel and Wings, remained un- 
changed at $6.40 and $6.38. Attempts to verify all of these data by visits to three 
Chicago wholesalers were unsuccessful. All were reluctant to give exact price in- 
formation. One frankly stated that the current asset value of his presence on the 
manufacturers' "direct lists" was at least $5,000 and expressed fear that, if the divul- 
gence of such confidential information were traced to him, he might be removed 
from their "direct lists." He therefore refused to talk. 

15 Thus, during the thirties, the list prices of Pall Mall, Tareyton, Philip Morris, 
and Raleigh frequently differed both in level and timing of price changes from 
the three major brands. These differences now appear to have almost wholly dis- 


panies been so dominant. 16 The two recent price changes suggest that 
American — now that it has again clearly established itself as the nation's 
largest cigarette manufacturer — has displaced Reynolds as the recognized 
price leader of the industry. It is even possible that Reynolds has volun- 
tarily abdicated its traditional position as price leader in order that it can 
be certain to avoid the absolute price identity so recently condemned. 
Despite the absence of any fundamental changes in their historical 
pattern of similar price behavior and despite their recrudescent degree 
of market control, the three convicted tobacco companies nevertheless 
appear to have followed a more moderate price policy since 1946 than 
that of earlier years. Net prices to the major manufacturers are now 
only about 33 per cent above 1939 and the Big Three's rates of profits 
are currently running at about 13-16 per cent (1948) as compared with 
14-18 per cent in 1939 and 17-22 per cent in 1931. (Meanwhile, Loril- 
lard's rate of profits has increased slightly to 10 per cent [1948] and 
Philip Aiorris' has fallen from 25.2 per cent in 1939 to 8.6 per cent in 
1947.) This more moderate price policy is probably in part due to the 
Big Three's desire to fend off the existing and potential competition 
which they found so unexpectedly strong in the thirties. Apparently 
such price policies — in conjunction with the continued powerful differ- 
ential advantage of their highly advertised brands — promise to maintain 
or increase the Big Three's market control where the more grasping 
price policies of the past failed to do so. Undoubtedly, present cigarette 
prices also still reflect to some extent the long period of controlled prices. 
Finally, because of the legal doubts which it cast upon the whole fabric 
of traditional cigarette price policies, recent antitrust action should prob- 
ably receive part of the credit for these more moderate prices. If so, how- 
ever, until effective remedial action is taken, the Antitrust Division's 

16 The following table summarizes the relevant data by companies: 

% Total Tax-Paid 
Company Cigarette Sales (Billions) Cigarette Production 

1931 1939 1947 1931 1939 1947 

American 46.1 41.1 121.6 39.4 22.7 36.1 


Liggett and Myers 

Philip Morris 


Brown & Williamson 



























































Total tax-paid production 







Big Three 







Economy brands 







The 1931 and 1939 data are compiled or computed from Record on Appeal, Exhibits No. 638, 702, 677, 428, 
293, 437, 1268, 1265, 1266. The 1947 data (domestic sales only) are estimated from Business Week, January 17, 
1948, p. 42. 


victory at best may have brought somewhat lower cigarette prices (at 
least in the short run) at the cost of that increasing concentration of 
economic power which it so much abhors. With this possible exception, 
the current history of the tobacco industry suggests that the widespread 
conviction of oligopolistic industries, though now legally possible, will 
produce only relatively unimportant economic consequences unless ac- 
companied by measures to change the underlying industrial structure 
which makes the condemned behavior almost inevitable. Thus the diffi- 
cult problem of evaluation and prescription in the public interest still 


The Antitrust Division has certainly not been unaware of the need 
for remedial action in the tobacco industry. In response to warnings 
from the Department of Justice following the 1946 decision, the con- 
victed companies claim to have made changes in their policies which 
eliminate the possibility of further violations of the antitrust laws. That 
the Antitrust Division has not yet been convinced is indicated by the 
fact that, during the last year, it has held several conferences with the 
three companies to consider the terms of a proposed consent decree. 17 
I am extremely skeptical, however, about the efficacy of any consent 
decree which falls short of dissolution. And even with the bargaining 
strength which its recent successful criminal action gives the Division, 
we can hardly expect the companies to agree to dissolution in such an 
out-of-court settlement. Yet in the civil suit recently instituted against 
the four major meat packers, the Division reveals that it considers 
dissolution the appropriate remedy for oligopoly. In this suit, patterned 
in detail after the Tobacco decision, the Division seeks dissolution of the 
four principal firms into fourteen companies. 18 Why, then, has not the 
Antitrust Division brought a civil suit seeking dissolution of the major 
tobacco companies? 

I suspect that the principal reason, apart from the need for economiz- 
ing resources, is that the Division recognizes the difficulties involved in 
formulating a dissolution plan for the tobacco industry. Each of the three 

17 R. J. Reynolds Tobacco Co., Prospectus on an Issue of Debentures, Septem- 
ber 29, 1948, pp. 27-28. 

18 In the Dist. Ct. of the U.S., Northern Dist. of 111., Eastern Division, U.S. v. 
Armour, Swift, Cudahy, and Wilson, Civil Action No. 48C 1351, September 15, 1948. 
The complaint alleges that, as a continuing offense since 1893, the four defendants 
have (1) refrained from competition among themselves by market sharing (constant 
percentages) in the purchase of livestock and the sale of meats, and by identical 
cost-figuring, prices, and terms of trade for both livestock and meats; (2) restrained 
competition by independents through the formulation of policies which the latter 
were urged to follow; and (3) excluded competitors by purchase or by resisting 
expansion of independents. As relief, the government asks ( 1 ) that each defendant 
be enjoined from following each of the practices complained of and (2) that Swift 
and Armour each be dissolved into five companies, and Cudahy and Wilson each be 
dissolved into two companies. 


companies has three to four cigarette manufacturing plants which could 
be made into separate firms. Furthermore, it is almost certain that the 
principal economies of scale, beyond those of the individual plant, are 
those of advertising and market control; hence are private rather than 
social. Nevertheless, there are at least two important barriers to dissolu- 
tion. First, each of the three firms has concentrated wholly (Reynolds) 
or largely (85 per cent or more) upon a single brand of cigarettes. Since 
a single brand of cigarettes could hardly be divided among several suc- 
cessor companies, the present major brands would probably have to be 
abolished. The original Tobacco Trust, having a multiplicity of brands, 
did not present this difficulty to those responsible for developing the 
details of its dissolution. Second, the paramount importance of advertising 
and sales effort in the industry cannot be overlooked. As Jones once 
pointed out, the dissolution of the Tobacco Trust "led to a duplication 
of selling organization and an increased overhead expense; and the re- 
sult was a general increase in selling costs." Between 1910 and 1913, the 
selling costs of the successor cigarette companies increased by 85 per 
cent, and the advertising expenditures more than doubled, in comparison 
with those of the Trust. 19 True, the dissolution ushered in a decade of 
innovation and price competition which was strongly in the public in- 
terest. Thereafter, however, the industry settled down into a pattern of 
non-price competition which it is doubtful that even a second dissolution 
could fully avoid. In other words, the cigarette industry is of such a 
nature that competition, at best, must continue to have important im- 
perfections. Nonetheless, despite such difficult problems as these, a work- 
able plan of dissolution could probably be devised upon the basis of a 
careful study of the structure of the tobacco industry. 

Whether or not dissolution is resorted to, however, supplementary 
techniques should not be ignored. Although lying outside of the limits 
of antitrust action, two other measures deserve consideration as means of 
encouraging competition in the tobacco industry. First, a sharply pro- 
gressive tax might be imposed upon the individual firm's total expendi- 
tures for advertising. Ideally, the tax should apply only to advertising 
expenditures greater than those already being made by firms of inter- 
mediate size, thereby permitting small firms to expand their advertising 
outlays somewhat but forcing the largest firms to curtail theirs con- 
siderably. Such a tax should go far toward eliminating the overwhelming 
advantage which large-scale advertising gives to giant firms, both in hold- 
ing their market position against existing small firms and in limiting the 
entry of new firms. If clear limits were established on the extent of ad- 
vertising, resort to price competition should also be encouraged. The 
great practical obstacles to this proposal are obvious. First, since such a 
tax might have to apply to all industries nondiscriminately, it would be 

Eliot Jones, The Trust Problem in the United States (1922), pp. 40-44. 


difficult to make due legislative allowances for important differences be- 
tween industries in the absolute and relative levels of advertising costs. 
Second, insofar as large firms lost business as a result of the lower ad- 
vertising expenditures caused by the tax, a question of constitutionality 
under the "due process" clause might be invoked. Third, one could hardly 
expect the press and radio to support such a tax with enthusiasm or com- 
plete objectivity. Nevertheless, this proposal does center attention upon 
one of the key problems involved in introducing a modicum of competi- 
tion into industries — such as tobacco, liquor, drugs, cosmetics, etc. — which 
rely heavily upon advertising. 

A second measure for encouraging competition in the tobacco indus- 
try would be the sharp reduction or elimination of the federal and state 
cigarette taxes. These taxes have now reached so high a proportion of the 
final retail price as to make price competition among cigarette manufac- 
turers almost prohibitive. Thus, at the present time, in a state levying a 
three-cent-per-package cigarette tax in addition to the federal tax of 
seven cents, a manufacturer would have to cut his final net price by 
about 15 per cent to lower the retail price by 5-6 per cent (or one cent 
a package). Since distributing margins are very small, the elimination of 
the cigarette tax would bring these two percentages very close together 
and make price competition among manufacturers much more attractive. 
Valuable though cigarette (and liquor) taxes may be as a lucrative source 
of governmental revenues, that they strongly foster monopolistic price 
policies has been almost wholly overlooked. It is true that, in the short 
run, the elimination of such taxes might simply increase the monopolv 
profits of the major firms. But the long-run result would probably be 
strong encouragement of entry and price competition, especially if com- 
bined with the advertising tax previously mentioned. If cigarette taxes 
are not eliminated, however, at least they should be based upon value 
rather than quantity of product. The present flat tax per thousand cig- 
arettes is inexcusably burdensome upon the economy brands, which are 
the only real element of price competition in the industry. While a grad- 
uated tax, based upon two or more price classes, has been frequently 
proposed, it has never met with Congressional favor. 20 Far better than 
such a graduated tax would be a straight ad valorem tax which would 
encourage a full continuum of possible prices, thereby helping to under- 
mine the present pattern of virtual price identity for both standard 
brands and economy brands. 

In spite of its apparent reticence to dissolve the major cigarette com- 
panies, the Antitrust Division may be expected to use dissolution pro- 

20 See, for example, Senate Committee on Finance (73d Cong., 2d sess.), Hearing 
on Reduction of Tax on Cigarettes, 1934. In December, 1948, the House Select Com- 
mittee on Small Business recommended to Congress "a graduated ad valorem rate of 
tax" on cigarettes. Report of House Select Committee on Small Bus'mess, Problems 
of S?Jiall Business Resulting from Monopolistic and Unfair Trade Practices (80th 
Cong., 2d sess.), pp. 10-11 and 26. 


ceedings much more frequently in the future than it has previously done. 
Unfortunately, drastic though it may be, dissolution appears to be the 
only really effective remedy for oligopoly which lies within the limits 
of antitrust action per se. If the current suit against the meat packers is 
indicative, the Division may even be launching an unprecedented drive 
toward atomization of American industry. The economic consequences 
of such remedial action, if generally executed, 21 would certainly be far- 
reaching. Whether they would always be in the public interest as well 
is less easy to foresee. The danger is that the well-known antibigness bias 
of the Antitrust Division will lead to an overzealous disregard for econo- 
mies of scale and other basic economic realities in at least some of its 
dissolution proposals. 


The Tobacco case is clearly a legal milestone in the social control of 
oligopoly. By permitting the inference of illegal conspiracy from de- 
tailed similarity of behavior and by shifting attention from the abuse of 
power to its mere existence (as indicated by degree of market control), 
the courts have at last brought oligopolistic industries within reach of 
successful prosecution under the antitrust laws. This is all to the good. 
The economic consequences will depend, however, upon whether gov- 
ernment victories are accompanied by appropriate remedial action. If 
thev are not, such victories will be nearly futile. On the other hand, if 
remedial action is taken, the Antitrust Division must assume a new and 
heavy responsibility to restrain the narrowly punitive spirit to which its 
antibigness bias so easily leads; and to prescribe remedies solely with a 
view to their contribution to the public interest, broadly conceived and 
based upon thoroughgoing economic analysis. Although the courts now 
may have largely abandoned the "rule of reason," the Antitrust Division 
— in deciding what industries to prosecute and in preparing appropriate 
corrective measures — must develop a "standard of reasonableness" of its 
own if the public interest is to be properly served. Finally, it must be 
recognized that other legislative reforms — though lying outside the 
bounds of antitrust action — can do much to supplement or complement 
the antitrust laws in attaining the goal of a more competitive economy. 

The Tobacco case is indeed a legal milestone. Whether it will be an 
economic milestone — or millstone — will depend upon the judiciousness 
with which its doctrines are applied. Certainly, the law and economics 
alike should combine their best efforts in meeting the new challenge 
which the Tobacco decision has so forcefully laid down. 

21 Even if the Division should meet with success in the courts, however, wide- 
spread dissolution proceedings (because of their drastic nature) would be not un- 
likely to lead to countermanding legislation by Congress. 

The Cellophane Case and the 
New Competition* 


On December 13, 1947 the Department of Justice instituted civil pro- 
ceedings against E. I. du Pont de Nemours & Company, charging du Pont 
with having monopolized, attempted to monopolize, and conspired to 
monopolize the manufacture and sale of cellophane and cellulose caps 
and bands in the United States in violation of section 2 of the Sherman 
Act. Almost precisely six years later Paul Leahy, Chief Judge of the 
United States District Court for the District of Delaware, rendered a 
decision in the matter. 1 He pointed out that the charge against du Pont 
of having monopolized cellophane involved two questions: "1. does du 
Pont possess monopoly powers; and 2., if so has it achieved such powers 
by 'monopolizing' within the meaning of the Act and under United 
States v. Aluminum Company of America [?]" He concluded that "un- 
less the first is decided against defendant, the second is not reached." 2 
Judge Leahy did not need to reach the second question for he found the 
defendant not guilty. In doing so he concluded that "[fjacts, in large 
part uncontested, demonstrate du Pont cellophane is sold under such in- 
tense competitive conditions acquisition of market control or monopoly 
power is a practical impossibility." 3 In reaching this conclusion Judge 
Leahy reviewed at length evidence introduced by the defendant to show 
that du Pont behaved like a competitor, not like a monopolist. The court 
found that du Pont conducted research to improve manufacturing effi- 
ciency, to reduce cost of production, and to improve the quality and de- 
velop new types of cellophane. It promoted the development and use of 

* The American Economic Review, Vol. XLV (1955), pp. 29-63. Reprinted by- 
courtesy of the publisher and the authors. 

t Vanderbilt University and the University of Wisconsin, respectively. 

1 United States v. E. I. du Pont de Nemours & Co., 118 F. Supp. 41 (D. Del. 
1953). This study is based largely on the testimony and exhibits in this case, but it 
does not consider cellulose caps and bands. Du Pont discontinued making caps be- 
fore the government filed its complaint, and the district court, as with cellophane, 
found no monopolizing of bands. The Supreme Court has indicated that it will re- 
view this decision. References to the government's exhibits will be designated as GX, 
to the defendant's exhibits as DX, and to the transcript of testimony as T. 

2 118F. Supp. at 54. 

3 Ibid., pp. 197-98. 



packaging machinery that could handle both cellophane and other flexi- 
ble wrapping materials. In doing so it not only helped to increase cello- 
phane sales but stimulated improvement in rival flexible wrapping ma- 
terials. It supplied customers with technical services to help them solve 
problems created by the use of cellophane. It developed over fifty types 
of cellophane tailored to meet the special wrapping needs of particular 
products. It studied the buying habits of the public. It conducted mar- 
ket studies to determine the effect on sales of packaging a product in 
cellophane. It promoted sales by educating potential cellophane users 
to the sales appeal of a transparent wrapping material. It reduced prices to 
get into new and broader markets. The court found that in response 
to price and quality changes buyers at times shifted from cellophane to 
competing products and back again. The court concluded that "[t]he 
record reflects not the dead hand of monopoly but rapidly declining 
prices, expanding production, intense competition stimulated by creative 
research, the development of new products and uses and other benefits 
of a free economy." 4 

This conclusion, based as it is on 7,500 pages of testimony and 7,000 
exhibits, cannot easily be dismissed. Many economists relying on the 
logic of the "new competition" will find in it support for their theories. 
One has already pronounced Judge Leahy's opinion a victory for our 

Such an optimistic conclusion so lightly reached by an economist and 
such high praise so extravagantly given by a judge warrant, first, a brief 
statement of the criteria by which economists can determine the exist- 
ence of monopoly and, second, an application of the relevant criteria to 
du Pont's cellophane operations in an effort to answer the question, has 
du Pont had monopoly power in making and selling cellophane? 

Detecting monopoly is simpler than measuring it. 5 While economists 
recognize that few if any industrial markets are free entirely from the 
influence of monopoly, by studying the structure and behavior of mar- 
kets they can generally isolate characteristics which taken together will 
permit them to classify markets as effectively competitive or noncom- 
petitive. In trying to classify du Pont's market for cellophane, we shall 
rely primarily on three criteria: (1) What role has business strategy 
played in du Pont's production and sales policies? (2) Is cellophane suf- 
ficiently differentiated from rival products to have a distinct market, or 
is its market that of all flexible wrapping materials? (3) Do the trend 
and level of its earnings reflect monopoly power or competition? 6 

4 Ibid., p. 233. 

5 Fritz Machlup is probably correct in concluding that "so many different ele- 
ments enter into what is called a monopolistic position and so complex are their 
combined effects that a measurement of 'the' degree of monopoly is even concep- 
tually impossible." The Political Economy of Monopoly (Baltimore, 1952), p. 527. 

6 Clair Wilcox uses the following criteria in classifying markets in his TNEC 
study, Competition and Monopoly in American Industry (1940): (1) the number 



Economists have said a good deal about the role which strategy plays 
among oligopolists jockeying for market position. They have said less 
about the significance of business strategy as a basis for classifying an 
industry as monopolistic or workably competitive. We believe it is an 
important criterion. Purely competitive markets do not generally con- 
front buyers and sellers in the business world. Frequently sellers are 
few, products are differentiated, knowledge is imperfect, obstacles to the 
movement of factors exist. Business firms from time to time make de- 
liberate adjustments in both their price and production policies; they 
resort to strategy to improve their lot. Strategy may be directed to other 
than price and production policies. Business executives are constantly 
alert for any business advantage that will make their market position 
more secure or isolate them from the impact of competitive forces. They 
seek control of the sources of the best raw materials and the richest 
natural resources. They try to improve their products and processes or 
to discover and develop new and better ones. They try to protect their 
accumulated know-how as business secrets or, where they can, to obtain 
patents that legalize monopoly. 

Economists recognize these practices as manifestations of business ri- 
valry, as aspects of the sort of competition that characterizes modern 
industrial markets. Business rivalry is itself a symptom of the absence of 
pure competition. Farmers who, lacking government aid, sell in competi- 
tive markets do not regard each other as business rivals but as neighbors. 
But even when businessmen forego active price competition, they gen- 
erally do not abandon all rivalry. Correctly, economists have concluded 
that this rivalry may protect the public interest. It leads to technological 
innovation and to economic progress. Although economists recognize 
that business strategy may lead to monopoly, some economists believe 
that in a dynamic capitalistic society monopoly is inevitably short-lived. 
It is continually being undermined by the rivalry of other firms. The 
better product, the better process of today gives way to the better prod- 
uct, the better process of tomorrow. Only the imperfections and mortal- 

of producers and the extent of industrial concentration, (2) uniformity of price 
quotations, (3) degree of price flexibility, (4) volume of production and extent of 
utilization of capacity, (5) rate of profit, and (6) rate of business mortality. Alone 
no one of these is a satisfactory index, and together they may be misleading unless 
perchance there is a consistency among the several indexes. We place considerable 
emphasis on two factors not included in Wilcox's list, business strategy and product 
differentiation, and we consider only incidentally if at all most of the factors on 
which Wilcox relied. Applying Wilcox's criteria to our conception of the cellophane 
market, we find that producers are few, concentration is high, profits are high, turn- 
over of producing units is low, business mortality is low. These criteria suggest 
monopoly power. On the other hand, cellophane prices have been flexible and sur- 
plus capacity has been negligible. These characteristics suggest competition. Whether 
or not the factors we have chosen are adequate to answer the question we have 
raised we leave to the reader. 


ity of monopoly make it tolerable. Businessmen striving for monopoly 
promote the public welfare by failing to achieve their goal. Where they 
achieve it, either by independent business strategy or by collusive action, 
the public interest may not be served. 

With these principles in mind, let us examine du Pont's strategy in 
developing the cellophane business in the United States. In doing this 
we do not mean to suggest that its strategy was immoral or unlawful. 
As Knauth has so well said: 

The contracts and arrangements which businessmen make from day to day 
seem to them wise, prudent, sound, and inherent in the nature of modern busi- 
ness. When their practices receive legislative interpretation and are denounced 
as monopolistic, they are puzzled. What has hitherto been deemed eminently 
proper and ethical now subjects them to unexpected criticism and opprobrium. 7 

No opprobrium is intended in this analysis. 

Du Pont became acquainted with cellophane through its production 
of artificial silk. In 1920 it had entered into a contract with the Comptoir 
des Textiles Artificiels, a French corporation, which through its affiliates 
was then an important manufacturer of rayon in France, Switzerland, 
Belgium, and Italy, for the joint operation of an American rayon com- 
pany using the viscose process. The viscose solution for making rayon 
was practically identical with that used in making cellophane. 8 The 
Comptoir had made about 970,000 pounds of cellophane in 1922, nearly 
40 per cent of which it sold in the American market as a transparent 
wrapping material. Aware of the affinity of rayon and cellophane proc- 
esses and impressed by the prospects of large cellophane sales in the 
American market, du Pont in 1923 signed an option contract with Arena 
Trading Corporation, 9 a Delaware corporation which was acting for it- 
self and its associates, including La Cellophane, Societe Anonyme, the 
Comptoir's affiliate which made cellophane. Under the option Arena pro- 
vided du Pont with all relevant economic and technical information to 
enable it to decide within four months whether it wished to make and 
sell cellophane in North and Central America through a corporation 
jointly owned by it and La Cellophane. If du Pont decided affirmatively, 
Arena agreed to transfer to the new corporation its technical knowledge, 
patent rights, trade marks, and good will and the exclusive right to make 
and sell cellophane in the North and Central American markets. This 
arrangement apparently contemplated the new corporation's becoming 
the sole producer of cellophane in these markets. At that time only the 
French Comptoir through its affiliates made cellophane any place in the 

On June 9, 1923 du Pont entered into an organizational agreement 

7 Oswald Knauth, Managerial Enterprise (New York, 1948), p. 11. 

8 Report of Dr. Fin Sparre, head of du Pont's development department, April 14, 
1923, GX 392, pp. 5431-32. 

9 Agreement of January 6, 1923, GX 1458, pp. 5999-6008. 


providing for the transfer to the new company, Du Pont Cellophane 
Company, Inc., of "an unqualified, unrestricted and exclusive right to 
use all and every process now owned" by Arena "or which may here- 
after be acquired by it ... in connection with the manufacture of 
cellophane." 10 

Before entering into this agreement du Pont had made an intensive 
study of the market possibilities of cellophane and of its production 
problems and decided that Arena could not deliver all the protection 
from competition that it had promised. About the patents du Pont's de- 
velopment department had said: "[T]he patent protection at present is 
exceedingly inadequate not to say worthless, and the future patent pro- 
tection is problematical because it is based on applications for patents, 
the issuance of which may not be determined for a matter of two years 
or more." 11 The development department had concluded that it was "by 
no means certain that the American Cellophane Company could maintain 
a monopoly on the strength of either present or prospective patents." 12 
Du Pont accordingly insisted on a provision in the organizational agree- 
ment that should the patent protection prove inadequate, Arena was to 
forfeit ten thousand shares of common stock in the new company. 13 

Du Pont apparently recognized that La Cellophane's trade secrets 
promised a protection from competition that the patents did not, and 
that as the first domestic producer of a differentiated if not unique prod- 
uct 14 it could for some time at any rate anticipate monopoly revenue in 
making and selling cellophane. It calculated that with an investment of 
$2,000,000, at current domestic prices for imported cellophane it could 
earn $631,832, an annual rate of 31.6 per cent. 15 This it regarded as suf- 
ficiently attractive to justify the venture. 

Du Pont the Sole Domestic Producer 

Du Pont became the sole domestic producer of cellophane and thereby 
a monopolist in its sale. The Department of Justice contended that it was 
an unlawful monopolist from the outset, 16 but the district court decided 

10 GX 1001, p. 992. The agreement specifically excluded processes which might 
subsequently be acquired by Arena from third parties, but it gave du Pont an 
option to purchase such rights for the North and Central American markets. Al- 
though Arena did not guarantee the validity of the patents to be licensed {ibid., 
p. 989), it specifically promised to give the new company "the exclusive right to 
manufacture cellophane in North and Central America to be used for any purpose 
whatsoever." Ibid., p. 993. Arena and Du Pont Cellophane signed their license agree- 
ment December 26, 1923. GX 1002, pp. 998-1001. 

11 GX 392, p. 5433. 

12 Ibid., p. 5434. Emphasis supplied. 

13 GX 1001, pp. 989-90. 

14 Du Pont's development department concluded that glassine, sheet gelatin, and 
tin foil, cellophane's closest rival products, offered no serious competition because 
of price or quality differences. GX 392, pp. 5437-38. 

15 Ibid., p. 5451. 

1G Brief for the United States of America, pp. 150-56, United States v. E. I. du 
font de Nemours & Co., 118 F. Supp. 41 (D. Del. 1953). 


otherwise. Whether lawful or not, du Pont was a monopolist in produc- 
ing cellophane, and it anticipated and in fact earned monopoly profits 
from the outset. 

This is a characteristic of any successful innovation. As Knight has 
pointed out: 

There is ... no clear distinction in practice between profit and monopoly 
gain. . . . New products . . . must also yield enough temporary monopoly 
revenue to make such activities attractive. 17 

But as Knight has also pointed out, we must distinguish between justifi- 
able monopoly revenue — returns to the innovator — and what Knight calls 
monopoly gains. Monopoly gains according to Knight are monopoly 
revenues that are "too large or last too long." What is too long or too 
large Knight does not say, but he clearly implies that the procedure by 
which they are made large and perpetuated may convert justifiable mo- 
nopoly revenue into socially unjustifiable monopoly gains. 

Having achieved at the outset a monopoly in producing and selling 
cellophane in the American market, du Pont took steps to protect its 
position. r 

One of its first strategic moves was to obtain an increase in the tariff. 
This became urgent in 1925, when Societe Industrielle de la Cellulose 
(SIDAC) completed a cellophane plant in Belgium and began exporting 
cellophane to the American market at cut-rate prices. Du Pont first con- 
sidered a patent infringement suit against Birn & Wachenheim, SIDAC's 
American distributors, but fearful that it would lose such a suit decided 
against it and in favor of a try for higher duties. 18 Its first step in getting 
the tariff raised was to request the United States Commissioner of Cus- 
toms to reclassify cellophane as a "cellulose compound" instead of as a 
"gelatin compound." When the Commissioner refused, du Pont appealed 
to the United States Customs Court. Its appeal was successful; the court 
ordered a reclassification and on February 24, 1929 the duty increased 
from 25 to 60 per cent ad valorem. 19 Apparently this was enough to pre- 

17 F. H. Knight, "An Appraisal of Economic Change — Discussion," Proceedings of 
the American Economic Association, American Economic Review, Vol. XLIV (May, 
1954), p. 65. 

18 About the suit Dr. Sparre of du Pont wrote W. C. Spruance, du Pont vice 
president, on August 3, 1925: "My belief is that it would cost a whole lot of money 
and that we would lose in the end, that is if the other side would be willing to 
fight." GX 1069, p. 1153. About du Pont's effort to get higher duties on cellophane, 
L. A. Yerkes, president of Du Pont Cellophane, had written Spruance on July 25: 
"In order that you shall be entirely familiar with the Cellophane status, I want to let 
you know that we are endeavoring to have the duty on Cellophane raised from 
25% to 45%, and Curie, Lane and Wallace are of the opinion that we have a fair 
chance of getting this through." GX 1068, p. 1142. 

19 The district court appropriately characterized du Pont's protest against what 
du Pont regarded as an improper classification of cellophane as "the normal act of a 
business concern engaged in active competition with importers of foreign products." 
118 F. Supp. at 167. The court also recognized that the tariff readjustment eventually 
shut out foreign competition. Ibid., p. 221. 


vent price cutting by importers. At any rate du Pont's quarterly com- 
petitive report for the second quarter of 1929 stated: 

The present tariff rate (.40 per pound) as fixed by the United States Cus- 
toms Court, has increased the cost of importing Transparent Cellulose Sheeting 
to such an extent that the competitors are adhering more rigidly to their pub- 
lished price list. Their selling policy in the past has been to obtain preference 
with the manufacturer by offering special price concessions. 20 

Du Pont won the field so completely from imported cellophane that 
its cellophane sales for 1929 represented 91.6 per cent of the total busi- 
ness in the United States, 21 whereas importers had had 21 per cent in 1927 
and 24 per cent in 1928. 22 The Tariff Act of 1930 fixed the duty on im- 
ported cellophane at 45 per cent ad valorem, 23 and cellophane imports 
were never again significant. In no year between 1930 and 1947 did they 
amount to 1 per cent of cellophane consumption in the United States. 24 

Division of World Markets 

La Cellophane's plan to develop the American market through a sin- 
gle company jointly owned by it and a domestic firm was not unique. 
Before transferring to Du Pont Cellophane Company, Inc., its rights to 
the American market, La Cellophane had made a similar agreement with 
Kalle & Company (hereinafter Kalle) covering the German market. Ul- 
timately Kalle obtained exclusive rights to La Cellophane's process and 
patents for the manufacture and sale of cellophane in Germany, Austria, 
Hungary, Czechoslovakia, Yugoslavia, Poland, Russia, Romania, China, 
Denmark, Sweden, Norway, and Finland. 25 

Although La Cellophane had agreed to furnish du Pont with such 
technological information and patent rights as it might later acquire from 
its other licensees, du Pont sought to fortify its market position through 
a direct agreement with Kalle. 26 On May 7, 1929 both parties agreed to 
exchange free of charge except for patent fees all patent rights and tech- 

20 GX 432, p. 5690. 

21 Du Pont Cellophane's quarterly competitive report, fourth quarter 1929, GX 

434, p. 5714. 

22 Ibid., first quarter 1929, GX 431, p. 5677. 

23 19 U.S.C.A. sec. 1001, par. 31(c). In 1951 the tariff was reduced to U 1 /? per 
cent ad valorem. United States v. E. I. du Pont de Nemours & Co., 118 F. 
Supp. 41, 167 (D.Del. 1953). 

24 GX 182A, p. 515A; GX 182, p. 515. 

25 Letter of October 30, 1929 from C. M. Albright, Du Pont Cellophane vice 
president, to the Buffalo office, GX 1091, p. 1195. 

20 As early as September, 1925, J. E. Crane, du Pont's European manager, had 
written H. G. Haskell, du Pont vice president, of talks with Dr. Duttcnhofer of 
Kalle & Company on the desirability of cooperation between du Pont and Kalle. 
Crane wrote: "Dr. Duttenhofer stated that as we were to cooperate in other mat- 
ters it would be a pity to compete in artificial silk and cellophane." GX 1393, 
p. 1800. 


nical data covering cellophane that they then had or might later get. 27 
This agreement did not specifically recognize a division of markets, but 
on October 30, 1929, C. M. Albright, Du Pont Cellophane vice president, 
listed for the Buffalo office the countries to which Kalle had exclusive 
rights. About the agreement Albright wrote: "The agreement, for ob- 
vious reasons, does not include the territorial limits, and it is suggested 
that this letter be attached to the copy of agreement for our future 
guidance." 28 By the summer of 1930 du Pont had patents on its moisture- 
proof cellophane in the United States and possessions, Belgium, France, 
and Italy and had applications pending in Great Britain, Canada, Japan, 
the European countries in Kalle's exclusive territory, and eight South 
American countries. 29 Du Pont assigned its moistureproof patent rights 
in the countries in Kalle's territory to Kalle or gave it implied licenses 
under which Kalle took out patents in its own name. 30 

Five years later du Pont entered a technical exchange and license 
agreement with British Cellophane Limited (hereinafter BCL), a La 
Cellophane licensee, which specifically delineated the territories within 
which each party would operate. 31 Under this agreement du Pont was to 
assign its British patents on moistureproof cellophane to BCL. Du Pont 
also assigned its French patents on moistureproof cellophane to La Cello- 
phane 32 and its Canadian patents to Canadian Industries Limited. 33 

Meanwhile all the world's leading cellophane producers except du 
Pont had tried to establish an international cartel to assign territories and 
fix quotas among themselves. Du Pont representatives attended the first 
day of the cartel conferences in Paris February 11-12, 1930 as "guests 
and observers" but did not sign the "official report" (agreement). 34 The 

27 GX 1087, pp. 1183-86. Perhaps another reason for du Pont's wishing to deal 
directly with Kalle is that on April 1, 1929 Du Pont Cellophane became du Pont's 
wholly owned subsidiary by an exchange of stock. Du Pont organized a new cor- 
poration, also named Du Pont Cellophane Co., Inc., La Cellophane to sit on its 
board as long as du Pont considered this in the new company's best interest. The 
1923 agreements remained in force. Agreement of March 18, 1929 between du Pont 
and La Cellophane, GX 1003, p. 1005. On July 1, 1936, du Pont dissolved Du 
Pont Cellophane Co., Inc., and replaced it with a cellophane division in its rayon 
department. Memorandum dated February 17, 1944, on the history of du Pont cello- 
phane, prepared in du Pont's cellophane division, GX 1, p. 8. 

28 GX 1091, p. 1195. 

29 Memorandum dated August 26, 1930, from Du Pont Cellophane's cellophane 
department to W. S. Carpenter, Jr., chairman of its board of directors, GX 2469, 
p. 3164. 

30 Du Pont Cellophane memorandum dated March 17, 1933, GX 1098, p. 1205; 
letter to Kalle dated March 20, 1933, GX 1099, p. 1206; memorandum dated April 27, 
1934, Review of the du Pont-Kalle Relations, prepared by du Pont's patent service, 
GX 1102, pp. 1210-12. 

31 Agreement of May 3, 1935, GX 1109, pp. 1229-34. 

32 GX 1102, pp. 1210-12. 

33 Letter dated February 12, 1942, transmitting patent assignments, GX 1187, 
pp. 1409-26. 

34 GX 1414, pp. 1841-44. 


district court found that they were not authorized to make commitments 
for du Pont and made none. Nevertheless the agreement recognized the 
North American market as belonging to du Pont and Sylvania. 35 It did 
not cover moistureproof or photographic cellophane. SIDAC and La 
Cellophane agreed to study the possibility of pooling their patents but 
with the understanding that this would not apply to du Pont patents. 36 

In 1934 du Pont relied on the 1930 cartel agreement in asserting its 
right to the West Indies as against BCL, to which La Cellophane had 
granted a license and with which du Pont was then negotiating its tech- 
nical agreement. In a December 13, 1934 letter to La Cellophane 37 du 
Pont referred to the minutes of the February 11-12, 1930 meetings of 
cellophane producers and quoted La Cellophane's letter of April 1, 1932 38 
describing the cartel's division of the world and containing the phrases, 
"Cuba being situated north of the Panama Canal, belonging thus to your 
territory." By May 21, 1935 the du Pont-BCL agreement 39 had been 
signed and du Pont and La Cellophane had agreed that du Pont's terri- 
tory included the West Indies except the possessions of European pow- 
ers. 40 

The cartel's course was not an easy one. World depression and the 
pressure of totalitarian governments for foreign exchange turned mem- 
bers' eyes toward South American markets, and even with agreements 
and quotas South American prices were unstable. Du Pont's sales there 
under its 1930 agreement with La Cellophane (discussed in the following 
subsection) were particularly disturbing to cartel members. 41 On Sep- 

35 Sylvania Industrial Corporation's entry into the American market is described 
in the following subsection. 

36 The court was not impressed by the "official report" of the cartel agreement. 
It stated: "Failure to prove any effect from it, this cartel aspect of the case raises a 
straight issue of fact. Du Pont did not make such an agreement." 118 F. Supp. 
at 221. Moreover, the court found that the actual conduct of the producers impli- 
cated in the agreement was inconsistent with the existence of a cartel. In the court's 
inimitable language, "Intricate theories of a conspiratorial network is cast aside." 
Loc. cit. 

37 GX 1034, pp. 1064-65. 

38 GX 1022, p. 1044. 

39 Agreement of May 3, 1935, GX 1109, pp. 1229-34. 

40 Memorandum dated May 21, 1935, "to be attached to original contract between 
E. I. du Pont de Nemours & Co. and Arena Trading Corporation of June 9, 1923," 
GX 1040, pp. 1075-76. 

41 Du Pont's 1930 agreement with La Cellophane gave du Pont half of whatever 
rights La Cellophane got in South America and Japan. Early in 1932 du Pont wrote 
La Cellophane that "we have made our prices in South America on plain Cellophane 
to correspond with yours" (letter dated January 29, 1932, GX 1112, p. 1248), and it 
tried to find out what quotas for South America La Cellophane and the other 
producers had agreed on; but it became dissatisfied with the operation of the agree- 
ment and on June 2, 1932 informed La Cellophane that henceforth du Pont would 
"consider ourselves free to pursue our own policy of sales in South America, Japan 
and China." GX 1023, p. 1045. La Cellophane did not take this declaration of inde- 
pendence too seriously and continued to point out that du Pont sold more than La 
Cellophane did in South America and Japan. On December 27, 1934 du Pont, al- 


tember 6, 1938 La Cellophane wrote du Pont that "it is apparently im- 
possible to bring about a price accord for South America in our Con- 
vention." 42 The second world war weakened still further agreements to 
divide markets. Du Pont's agreements with Kalle and BCL were to run 
twenty years, subject to renewal, but in 1940 du Pont disavowed all 
formal territorial limitations, not only with these companies but with 
Canadian Industries Limited and La Cellophane as well, "in the light of 
legal developments in this country." 43 

SID AC Competes with La Cellophane and with du Pont 

Although La Cellophane had promised du Pont a monopoly in making 
and selling cellophane in the United States, it could not fulfill the prom- 
ise. As du Pont feared, neither its patents nor its know-how was sufficient 
to protect it from competition. In 1925 two former employees of La 
Cellophane, using La Cellophane's trade secrets, helped establish SID AC, 
which began to sell in the rich American market. 44 It made its first sales 
through Birn & Wachenheim, who had handled La Cellophane's business 
in the United States before the organization of Du Pont Cellophane. In 
1929 SID AC established an American subsidiary, the Sylvania Industrial 
Corporation of America, and quit exporting cellophane to the United 
States. 45 By this time it had subsidiaries in England and Italy and com- 
peted in La Cellophane's export markets. La Cellophane sued SIDAC 
for patent infringement and in settlement accepted a stock interest in 
SIDAC; thus indirectly it became through Sylvania du Pont's competitor 
in the American market, in violation of its 1923 agreements with du Pont. 
Negotiations over this matter were prolonged. Du Pont conceived its 
problem to be how to "accept reparations and at the same time protect 
its future position without contravening American statutes." 46 In lieu of 
reparations La Cellophane lifted the 1923 restriction limiting du Pont to 
the North and Central American markets. La Cellophane granted it equal 
rights with itself in Japan and South America. 47 La Cellophane also agreed 
to keep technical information, patents, and other data which it received 
from du Pont from going directly or indirectly to SIDAC or Sylvania. 

though reiterating its stand taken in 1932, expressed willingness to exchange figures 
on its sales in those markets for La Cellophane's and those "of the other members." 
GX 1037, p. 1070. 

42 GX 1445, p. 1920. 

43 Identical letters dated October 17, 1940, GX 1273, p. 1602, GX 1274, p. 1603, 
GX 1275, p. 1604, GX 1276, p. 1605. 

44 Memorandum dated February 17, 1944 on the history of du Pont cellophane, 
prepared in du Pont's cellophane division, GX 1, p. 12. 

45 Du Pont Cellophane's quarterly competitive report, third quarter 1929, GX 433, 
p. 5702. 

46 Memorandum of a November 14, 1929 discussion by du Pont officials, GX 1410, 
p. 1831. 

47 Letter dated March 6, 1930 from du Pont to La Cellophane, GX 1013, pp. 
1027-29; excerpt from minutes of May 8, 1930 meeting of du Pont's board of di- 
rectors, GX 1015, p. 1031. 


Du Pont Seeks Patent Protection 

When du Pont obtained its option to participate jointly with La Cello- 
phane in developing the American market, it had not investigated the 
validity of La Cellophane's patent claims. The terms of the option had 
been "predicated on the practical absence of serious competition on the 
part of other manufacturers either in this country or other countries." 48 
Shortly after its organization Du Pont Cellophane launched a research 
program designed to strengthen its market position by improving cello- 
phane. One of its chief defects was its permeability to moisture. Du Pont 
promptly attacked this problem and by 1927 had developed a moisture- 
proofing process and had applied for patents. Its basic patent covering 
moistureproof cellophane, Charch and Prindle patent No. 1,737,187 is- 
sued in 1929, was a product patent broad in scope and extensive in 
claims. 49 J. E. Hatt, general manager of Du Pont Cellophane's cellophane 
department, in summarizing du Pont's moistureproof cellophane patent 
situation in 1930 recognized its vulnerability and indicated that du Pont 
had taken steps to bulwark it. He described patent applications that du 
Pont had filed and quoted patent counsel's opinion that they promised 
"important and substantial additional protection." 50 Between 1930 and 
1934 Du Pont Cellophane authorized a research project further to bolster 
its patent position. In reporting on the success of this project in 1934 
President Yerkes said: 

This work was undertaken as a defensive program in connection with pro- 
tecting broadly by patents the field of moistureproofing agents other than 
waxes which was the only class of material disclosed in our original Cellophane 
moistureproofing patents. 

The investigations on this subject did, in fact, lead to the discovery of a 
number of classes of materials which could serve equally well for moisture- 
proofiing agents. . . . Each of these classes has been made the subject of a 
patent. . . . Altogether, 13 patent applications are being written as a result of 
the work done under this project, all in view of strengthening our Moisture- 
proof Cellophane patent situation. 51 

48 Report dated April 14, 1923 by Dr. Fin Sparre, director of du Pont's develop- 
ment department, GX 392, p. 5455. See also pp. 5453-56. 

49 A problem arising during the second world war when the government needed 
more moistureproof laminated products than du Pont could supply directly, reflects 
the breadth of the patent claims. Hines of du Pont posed the problem in this way: 
"What is the best procedure to give the Government these laminated products nec- 
essary to win the war and, having decided on that, what can be done to preserve 
du Pont's position in a postwar economy?'' Recognizing that the government's 
interest might best be served by allowing converters to make them, du Pont feared 
that the converters might "at the end of the war, be possessed of a great deal of 
information with respect to the preparation of moistureproofing compositions and 
the technique of moistureproofing film with them and would be disposed to continue 
in such a business on a peace-time basis to the detriment of the Company's in- 
terests." Memorandum dated January 26, 1942 from du Pont's patent service to du 
Pont's cellophane research section, GX 2497, pp. 3255-57. 

50 Memorandum of August 26, 1930, GX 2469, p. 3160. 

51 December 1933 report to Du Pont Cellophane's board of directors, January 22, 
1934, GX 488, p. 6478. Du Pont spent $19,503 on this research project. This compares 


These steps proved adequate to forestall other domestic competition 52 
and to bring Sylvania Industrial Corporation to terms when it invaded 
the American market. 

Sylvania Reaches Accord "with du Pont 

Sylvania completed its Virginia plant for making cellophane in 1930. 
Apparently its early experimental research to develop a moistureproof 
cellophane rested, as did du Pont's, on the use of a nitrocellulose base 
to which gum, wax, and plasticizer were added. When du Pont's Charch 
and Prindle patent covering moistureproof cellophane was issued, du 
Pont advised Sylvania informally of its claims and Sylvania after con- 
sidering them "felt obliged to discard the work they had done up to 
that time, and approach the subject from a new angle." 53 Their new angle 
substituted a vinyl resin base for the nitrocellulose base. Du Pont, re- 
garding this as an infringement, advised Dr. Wallach, Sylvania's presi- 
dent, that "we would be obliged to enforce our patent" 54 and eventually 
filed an infringement suit against Sylvania. 55 In the antitrust proceedings 
against du Pont the government contended that the "entire infringement 
suit was nothing more than a harassing action designed to coerce Syl- 
vania into entering a highly restrictive agreement." 56 The district court 
in finding for du Pont rejected this contention. 57 Since the court has 

with an expenditure of only between $5,000 and $10,000 authorized in October 
1924 to hire a single chemist to develop the original moistureprooflng process. 
DX 393 and DX 394. Total expenditures for "technical activities expenses," which 
included all types of technical work designed to improve cellophane production 
and processes, came to only $32,048 during 1925 and 1926. DX 387. 

52 Du Pont's strong patent position may not have been wholly responsible for the 
reluctance of other domestic companies to produce cellophane. Apparently Union 
Carbide & Carbon Corporation in the 1930's considered entering the cellophane 
field. It purchased rights to a process for making a transparent wrapping material 
similar to cellophane. Lammot du Pont in a letter of December 2, 1931 to L. A. 
Yerkes, president of Du Pont Cellophane, stated that in the course of an hour's 
conversation on this topic with Messrs. Jesse Ricks and Barrett of Union Carbide & 
Carbon "[t]hey assured me repeatedly they did not wish to rush into anything, most 
of all a competitive situation with du Pont. Their whole tone was most agreeable. 
... In the course of the conversation, various efforts at co-operation between Car- 
bide and du Pont were referred to, and in every case assurances of their desire to 
work together, given." GX 4381, p. 4300. 

53 Memorandum dated February 18, 1931 from J. E. Hatt, Du Pont Cellophane's 
general manager, to its executive committee, GX 2482, p. 3204. 

54 Loc. cit. 

55 The bill of complaint in Du Pont Cellophane Company v. Sylvania Industrial 
Corporation appears in the record of the Cellophane case as GX 2479, pp. 3183-90, 
and Sylvania's answer as GX 2480, pp. 3191-99. 

56 Statement to the court by J. L. Minicus, counsel for the government, T. 2472. 

57 Judge Leahy said: "Neither party dictated the terms of the license agreement 
by which the suit was settled." 118 F. Supp. at 151. He based this finding on 
testimony by L. A. Yerkes, president of Du Pont Cellophane. Although Judge 
Leahy had said that if he found that du Pont did not possess monopoly power 
it would be unnecessary for him to pass on whether it had monopolized cellophane 
under section 2 of the Sherman Act and the principles of United States v. Aluminum 
Co. of America, 148 F. 2d 416 (2d Cir. 1945), he nevertheless made a decision on 


spoken, we do not express judgment on this issue. But we wish to review 
briefly evidence that throws some light on du Pont's strategy. 

The record indicates that (1) du Pont in negotiating for reparations 
following SIDAC's entry into the American market considered and re- 
jected a proposal that it grant Sylvania a license which would restrict 
its output; (2) after warning Sylvania that it would defend its patents 
and learning that Sylvania challenged their validity, du Pont postponed 
action while entrenching its patent position; 58 (3) although professing 
confidence in its ability to establish its patents' validity, du Pont offered 
to settle the issue by granting a license limiting Sylvania's production 
of moistureproof cellophane to 10 per cent of the companies' combined 
output; 59 (4) on Sylvania's rejecting this offer du Pont formally notified 
Sylvania that it was infringing du Pont's moistureproofing patents and 
asked that it cease; 60 (5) upon its refusal to desist du Pont formally in- 
augurated infringement proceedings; and (6) before the proceedings 
were carried to completion du Pont and Sylvania settled the suit by a 
patent exchange and licensing agreement. 61 

Both parties no doubt thought that they stood to gain by a settle- 
ment. If Sylvania lost the suit, it would be forced to stop producing 
moistureproof cellophane or to produce it on such terms as du Pont 
might offer. If it won, anyone with adequate resources could produce 
cellophane, and selling cellophane would become a competitive enter- 
prise. After a discussion with Sylvania's general counsel du Pont's patent 
attorney summed up Sylvania's plight as follows: 

During the conference Mr. Menken stated that in his opinion the case should 
be settled. He said that they were very fearful of what the result would be to 
their company in the event they succeeded in having the claims of the patents 

that charge also. He ruled that du Pont's licensing and technology exchange agree- 
ments with La Cellophane did not unreasonably restrain trade and that their ter- 
ritorial limitations were ancillary to the acquisition of trade secrets. 118 F. Supp. 
at 219. He ruled that the circumstances under which du Pont acquired its patents 
failed to show that the "acquisitions affected its ability to exclude competition" 
(ibid., p. 212), and that du Pont placed only "lawful and reasonable limitations on 
use" in its licenses (ibid., p. 211). In any event, he ruled, du Pont had a lawful 
monopoly in its moistureproof cellophane patent. He said: "Evidence does not dis- 
close combining of competing or independent process patents or efforts to control 
unpatented products" (ibid., p. 214). 

58 A running memorandum of developments in the du Pont-Sylvania patent 
controversy between July 9, 1931 and April 6, 1933, in the du Pont files, contains 
the following statement: "At Board meeting on 8/21/31 . . . [i]t was felt by Air. 
Pritchard [du Pont's patent counsel] that actual suit against Sylvania should not be 
instituted until we have these claims issued in form of actual patents. . . . LA.Y. 
[Yerkes, Du Pont Cellophane's president] still felt it would be desirable for us to 
have Sylvania under a license agreement if possible." GX 2478, p. 3181. 

59 Loc. cit. On August 27, 1931 Yerkes wrote to Dr. Wallach confirming an oral 
offer made on or about July 9, 1931. A memorandum dated July 13, 1931 outlines 
the terms of the offer. GX 2483, p. 3206. 

60 GX 2478, p. 3181. The memorandum refers to a letter dated 11/19/31 which 
lists the patents du Pont claimed to be infringed. 

61 Agreement dated April 26, 1933, GX 2487, pp. 3212-33. 


which are involved in the litigation held invalid. He seemed to realize the old 
adage that the defendent can never win. ... If the Du Pont Cellophane Com- 
pany succeeds and the patents are held to be infringed, Sylvania Industrial 
Corporation will be under injunction and will be obliged to stop manufacturing 
moistureproof wrapping tissue. On the other hand, if they succeed in having 
the broad claims of the patents held invalid they will throw the art open, as 
far as the broad claims are concerned, to anyone and therefore will have addi- 
tional competition. Sylvania . . . has plenty of ready cash but are hesitant 
about enlarging their plant facilities pending the litigation since, if successful, 
they will only invite further competition. 62 

With neither side ready to test the validity of du Pont's patents, the 
parties compromised. The compromise constituted no threat to du Pont's 
dominant market position. 

Under the settlement reached April 26, 1933, Du Pont Cellophane 
granted Sylvania a nonexclusive license (made exclusive in 1938) of du 
Pont's five basic patents on moistureproof cellophane and agreed to 
license to it any patents within their scope which du Pont might get 
before October 16, 1948. Sylvania agreed to grant similar rights to Du 
Pont Cellophane under any patents which it might get. Sylvania agreed 
to pay du Pont a royalty of 2 per cent of its net cellophane sales for the 
use of du Pont's basic patents and an additional 2 per cent if Sylvania 
accepted licenses under future du Pont patents representing departures 
from the five basic patents. But the settlement went further than a mere 
cross-licensing of present and future patents. It provided that Sylvania's 
production be restricted to a fixed percentage of total moistureproof 
cellophane sales, beginning with 20 per cent in 1933 and increasing by 
1 per cent until it reached 29 per cent in 1942. Should Sylvania exceed 
its share in a given year, it agreed to pay a penalty royalty of 20 cents 
a pound or 30 per cent of its net cellophane sales, whichever was higher. 
If du Pont used any of Sylvania's patents, it agreed to a similar penalty 
for exceeding its basic quota. But it never used them. 

Until June, 1951 du Pont and Sylvania were the only producers of 
cellophane in the American market. 63 Between 1933 and 1945 (when 
they contracted for smaller royalties and abandoned penalites for ex- 
ceeding their quotas), with Sylvania's output geared to du Pont's, du 
Pont could determine how much cellophane should come on the market. 
Actually the penalty provision of the agreement never operated and its 
deletion from the 1945 agreement produced no marked effect on Syl- 
vania's production. The court found that "[i]ts policies as to expansion 
in no way changed following the termination of the 1933 agreement in 
1945." 64 Although their shares varied from time to time, du Pont supplied 

62 Letter dated August 4, 1932 from W. S. Pritchard to B. M. May, GX 2811, 
pp. 6073-74. 

63 In June, 1951 Olin Industries, Inc., began the production of cellophane at 
Pisgah Forest, North Carolina. Testimony of Fred Olsen, Olin vice president, 
T. 6829. 

64 118 F. Supp. at 157. 


about 16 per cent and Sylvania 24 per cent of the market from 1933 to 
1950. 65 But gearing Sylvania's production to du Pont's must have lessened 
Sylvania's incentive to independent, vigorous rivalry, price or nonprice, 
and the record indicates that until January 1, 1947 Sylvania's quoted 
prices were generally identical with du Pont's. 66 


Du Pont's moves and countermoves to protect its domestic market 
were the strategy of a producer operating in a monopolistic, not a com- 
petitive, market. Its agreements with foreign producers to license patents 
and exchange technical data, its domestic patent program, its effort to 
get higher tariffs, its restrictive market agreement with Sylvania, all re- 
flect du Pont's effort to preserve what it apparently regarded as a mo- 
nopoly market. That du Pont and Sylvania (whose production was 
geared to du Pont's and whose quoted prices were generally identical 
with du Pont's) together monopolized the market for cellophane seems 
scarcely debatable. That du Pont acted as though in its monopoly of 
cellophane it had a valuable property right which it sought to exploit is 
equally clear. But was du Pont mistaken? Were available substitutes so 
similar that du Pont's monopoly of cellophane was in reality a mirage or 
a phantasy? Is there in fact no distinct market for cellophane, but only 
a larger market for flexible wrapping materials with producers so nu- 
merous that none can make monopoly profits? Let us turn to that question. 


For several years du Pont was the sole domestic producer of cello- 
phane and for a quarter of a century Sylvania and du Pont were the only 
producers. But buyers of flexible wrapping material need not rely solely 
on these two suppliers. Several hundred rivals produced flexible wrap- 
ping materials, in many uses substitutes for cellophane. May not these 

65 Data on production, 1933 to 1950, table in United States v. E. 1. du Pont de 
Nemours & Co., 118 F. Supp. 41, 116 (D. Del. 1953). In the five years following the 
expiration of the 1933 agreement Sylvania's percentage of total domestic production 
was only 1 per cent higher than its percentage in the five years preceding the 
expiration of the agreement. American Viscose Corporation acquired Sylvania in 

6G Du Pont and Sylvania not only quoted identical prices for their most important 
cellophane types, but their price changes almost always became effective on the same 
date. GX 549, pp. 7128-66. During the postwar period of short supplies and after the 
government had instituted its suit against du Pont, differences in Sylvania's and du 
Pont's prices appeared. DX 591, p. 1128. Judge Leahy was impressed not by the 
identity of quoted prices but by the fact that Sylvania at times made discounts 
from its list prices which du Pont did not match. In speaking of du Pont-Sylvania 
competition he declared that Sylvania "has continued to expand to the full extent of 
its financial resources" (118 F. Supp. at 212); and that although du Pont was 
superior in the services rendered to customers, in technology, in price, and in the 
development of special types of films, competition between the two companies has 
"flourished" (loc.cit.). 


have converted a monopolistic market into one of workable competi- 
tion? Let us examine briefly the relevant theory and then the facts. 

Price Theory and Product Differentiation 

Although others have made important contributions to an understand- 
ing of the significance of interproduct competition, Chamberlin, the 
pioneer, offers a good starting point for this discussion. Chamberlin has 
recognized that "[a]s long as the substitutes are to any degree imperfect, 
he [the seller] still has a monopoly of his own product and control over 
its price within the limits imposed upon any monopolist — those of the 
demand." 67 But Chamberlin also recognized that rival products, where 
entry is free and differentiation not marked, could eliminate excess prof- 
its even in the "monopolized" field. Expressing his findings diagrammat- 
ically, he concluded that the sloping demand curve facing the producer 
of a differentiated product may become tangent to the cost curve some- 
where above lowest average cost. Chamberlin regarded this as a "sort of 
ideal" solution. As he put it, "With fewer establishments, larger scales of 
production, and lower prices it would always be true that buyers would 
be willing to pay more than it would cost to give them a greater diversity 
of product; and conversely, with more producers and smaller scales of 
production, the higher prices they would pay would be more than such 
gains were worth." 68 

Chamberlin's conclusion that the entry of producers of substitute 
products will eliminate monopoly profits is based upon two important 
assumptions: (1) his uniformity assumption — "both demand and cost 
curves for all the 'products' are uniform throughout the group"; 69 and 
(2) his symmetry assumption 70 — "any adjustment of price or of 'product' 
by a single producer spreads its influence over so many of his competitors 
that the impact felt by any one is negligible and does not lead him to any 
readjustment of his own situation." 71 

If cost and demand curves are not uniform, of if the "group" of firms 
producing the substitute products is sufficiently small to introduce the 

67 E. H. Chamberlin, The Theory of Monopolistic Competition (5th ed., Cam- 
bridge, Mass., 1947), p. 67. 

68 Ibid., p. 94. This assumes, of course, that buyers know what they get and get 
what they want in buying a differentiated product. This is a dubious assumption. 
Years ago a well-known pharmaceutical company by its advertising endeavored to 
create a widespread fear of halitosis. "Not even your best friends will tell you." 
Having created a fear of halitosis, it provided a product to dissipate it, thereby 
rendering the buyer a service for which he was willing to pay. 

69 Ibid., p. 82. To simplify his exposition Chamberlin first assumes uniformity in 
cost and demand curves. Later he abandons this assumption in the interest of reality. 
In abandoning it he reaches the conclusion indicated in the text: where sufficientlv 
effective substitutes are not offered in the market, monopoly profits result. 

70 G. J. Stigler so describes this assumption. Five Lectures on Economic Proble?ns 
(London, 1949), p. 17. 

71 Chamberlin, op. cit., p. 83. 


oligopoly problem, we may expect a divergence from the above solution. 
As for the uniformity assumption, Chamberlin says: "[I]n so far as sub- 
stitutes of such a degree of effectiveness may not be produced, the con- 
clusions are different — demand curves will lie to the right of the point 
of tangency with cost curves, and profits will be correspondingly higher. 
This is the explanation of all monopoly profits, of whatever sort." 72 Thus, 
unless effective substitutes exist, Chamberlin argues that monopoly prof- 
its may be "scattered throughout the group." 73 If Chamberlin's symmetry 
assumption is not fulfilled, an oligopoly solution may be expected. 74 In 
either case monopoly profits result. 

In applying Chamberlin's theory to the flexible packaging materials 
market and to cellophane's position in it, the empirical issue revolves 
about (1) the degree of effectiveness of substitutes and (2) the number 
of rival firms. If substitutes are not effective enough to eliminate mo- 
nopoly profits, it is not necessary to consider the oligopoly problem. 

Clark's analysis 75 leads to similar conclusions, viz., that competition 
among substitutes may eliminate monopoly profits; but Clark goes fur- 
ther than Chamberlin in finding these results salutary. According to 
Clark the high cross elasticity of demand tends to flatten the monopolist's 
demand curve. Moreover, the monopolist's fear of potential competition 
may lead him to behave as though potential competition had become a 
reality. These two restraining forces, rival substitute products and po- 
tential competition, may yield cost-price relationships similar to those of 
pure competition. They may make imperfect competition workable. 

An increasing number of economists have come to believe this. Robert- 
son develops the idea somewhat further. In reviewing the significance of 
interproduct and interindustry competition he concludes that we really 

72 Ibid., p. 111. Emphasis in original. This statement of the problem seems to make 
it similar to if not identical with the conventional, neoclassical conception of mo- 
nopoly. Richard T. Ely for example pointed out: "The use of substitutes is con- 
sistent with monopoly, and we nearly always have them. For almost anything we 
can think of, there is some sort of a substitute more or less perfect, and the use 
of substitutes furnishes one of the limits to the power of the monopolist. In the 
consideration of monopoly we have to ask, what are the substitutes, and how 
effective are they?" Monopolies and Trusts (New York, 1912), pp. 35-36. 

73 Chamberlin, op. cit., p. 113. By the "group" Chamberlin apparently means firms 
making products which although differentiated are designed for the same use, e.g., 
toothpaste manufacturers. In his "Monopolistic Competition Revisited," Economica, 
Nov., 1951, N.S. XVIII, 352, 353, he abandons the group concept, arguing that "com- 
petition is always a matter of substitutes, and . . . substitutes are always a matter of 
degree." In abandoning the group concept he does not abandon the conclusion that 
where substitutes are similar enough and entry is free, monopoly profits will dis- 
appear and the demand curve will be tangent to the cost curve at some point 
above minimum cost. But he also recognizes that the "isolated" monopolist, in 
spite of close substitutes, may find the demand for his own product strong enough 
to yield him "profits in excess of the minimum." 

74 Monopolistic Competition, p. 102. 

70 J. M. Clark, "Toward a Concept of Workable Competition," American Eco- 
nomic Review, Vol. XXX (June, 1940), pp. 241-56. 


need not worry about monopoly for "there is probably not much of it." 
There is not much of it because the "old-fashioned apparatus of com- 
petition works in new ways to save us." 76 

Moreover, this new apparatus of competition once more makes rele- 
vant a theory of competition based on large numbers. 

To assess the competitive situation of a firm we must still resort to counting 
numbers. We cannot do away with the group, for the group exists in the real 
world. Yet counting only those firms which are within the "industry" tells us 
very little. We must do our counting by taking categories of uses for the out- 
put of an industry, considering what products of other industries directly com- 
pete within these categories. 77 

Since a monopolist's product may serve in a great variety of uses, a 
monopolist may find it "profitable to forego monopoly control in one 
use in order to push the commodity into many uses." 78 Thus monopoly 
serves the public by serving itself and in doing so loses its power over 
the market. 

What Robertson has discovered for the economists, businessmen had 
already professed. David Lilienthal, writing about the "new competi- 
tion," said: 

I am not saying that active competition between the producers of the same 
product is of no present consequence. It certainly is. My point is that under 
present-day conditions it is often the least significant form. The competition 
between alternative materials, or ways of satisfying human needs and desires, 
has become a new dimension of competition. 79 

It was on such principles that Judge Leahy relied in reaching his con- 
clusions in the Cellophane case. 

This calls for a more careful consideration of the uniqueness of cello- 
phane, of du Pont's pricing policies in selling it, and of the rate of earn- 
ings realized in doing so. If cellophane is sufficiently differentiated from 
other flexible wrapping materials, its demand curve may "lie to the right 
of the point of tangency with its cost curve" and its producer may re- 
ceive monopoly profits in making and selling it. If cellophane is a less 
highly differentiated product within Chamberlin's conception of the 
term and if entry to the manufacture of rival wrapping materials is not 
blocked, the maker of cellophane will be faced by a sloping demand 
curve; but the curve will be tangent to the cost curve at some point 
above lowest average cost, and the seller will not make a monopoly 
profit. If the differentiation is so slight and potential competition so im- 
minent as to bring it within Clark's concept of the term, the seller's long- 

76 R. M. Robertson, "On the Changing Apparatus of Competition," American 
Economic Review, Vol. XLIV (Proceedings of the American Economic Association, 
May, 1954), p. 61. 

77 Ibid., pp. 53-54. 

78 Ibid., p. 57. 

79 D. E. Lilienthal, Big Business: A New Era (New York, 1953), p. 60. 


run demand curve will be close to the horizontal (his control over price 
will be slight) and prices will be close to lowest average cost. If the cello- 
phane market conforms to Robertson's model, cellophane's differentia- 
tion will be too slight to count, monopoly profit will not exist, and its 
price will be competitive. To which of these models does the market 
for cellophane conform? 

The Market for Cellophane 

As a first step in answering this question we will examine briefly the 
flexible packaging materials market. The district court in determining 
whether du Pont monopolized the market for cellophane concluded that 
"the relevant market for determining the extent of du Pont's market 
control is the market for flexible packaging materials." 80 In this broad 
market the court found several hundred firms selling a variety of dif- 
ferentiated products for an even wider variety of uses. They sold either 
directly to packagers or to converters who prepared packaging materials 
for special uses. The court found that in 1949 du Pont cellophane ac- 
counted for only 17.9 per cent of the total square yardage of domestic 
output and imports of flexible packaging materials. 81 (Apparently this 
did not include kraft paper.) Such a small percentage scarcely demon- 
strates that du Pont had monopolized the flexible packaging materials 
market. Nor had it. But in passing judgment on the validity of the court's 
view that there is a single market for flexible packaging materials it may 
be helpful to classify the major contemporary materials according to 
their special qualities and major uses. 

Cellophane is a thin, transparent, nonfibrous film of regenerated cellu- 
lose. It comes in two major types: plain and moistureproof. Moisture- 
proof cellophane far outsells plain. In 1950 plain cellophane sales totalled 
$12,005,737; moistureproof cellophane sales, $ 116, 660,209. 82 Because mois- 
tureproof cellophane sales are over nine times those of plain, our analysis 
will give primary consideration to moistureproof. Moistureproof cello- 
phane is highly transparent, tears readily but has high bursting strength, 
is highly impervious to moisture and gases, and is resistant to grease and 
oils. Heat sealable, printable, and adapted to use on wrapping machines, 
it makes an excellent packaging material for both display and protection 
of commodities. 

Other flexible wrapping materials fall into four major categories: 
(1) opaque nonmoistureproof wrapping paper designed primarily for 
convenience and protection in handling packages; (2) moistureproof 
films of varying degrees of transparency designed primarily either to 
protect, or to display and protect, the products they encompass; (3) non- 

80 118 F. Supp. at 60. 

81 Ibid., p. 111. 

82 Tabic showing comparison of du Pont and Sylvania plain and moistureproof 
cellophane sales, 1924-1950, 118 F. Supp. at 123. 


moistureproof transparent films designed primarily to display and to 
some extent protect, but which obviously do a poor protecting job where 
exclusion or retention of moisture is important; and (4) moistureproof 
materials other than films of varying degrees of transparency (foils and 
paper products) designed to protect and display. 

Kraft paper is the leading opaque nonmoistureproof wrapping paper. 
For general wrapping it has no equal. It is cheap, strong, and pliable and 
gives adequate protection. On a tonnage basis it easily tops all other 
packaging materials in total sales. But it is neither designed for nor 
adapted to the special uses for which cellophane was created and, as one 
market expert has put it, "in the true sense" does not compete with cello- 
phane. More accurately, we think, cellophane does not compete with 
it. On a cost basis it cannot compete. At less than one cent per thousand 
square inches, kraft paper sells for less than cellophane's manufacturing 

The leading moistureproof films which might compete with cellophane 
include polyethylene, Saran, and Pliofilm. Relatively these are newcomers 
in the packaging field. In some qualities they match or even excel cello- 
phane. But we have it on the authority of du Pont market analysts that 
these films have offered little or no competition to cellophane in its major 
markets. According to du Pont's 1948 market analysis, prepared by its 
experts for company use in making decisions, although Saran was "su- 
perior in moisture protection, no significant commercial uses" had de- 
veloped for it "due principally to its high price" and "no substantial cost 
reduction" was in sight. 83 In 1949 a thousand square inches of 100-gauge 
Saran #517 sold for about 2% times as much as the same amount of mois- 
tureproof cellophane (see Table 1). Du Pont experts found polyethylene 
lacking in transparency, "too limp to operate satisfactorily on wrapping 
machines, . . . difficult to heat seal, print and glue," with "poor surface 
slip and high static, and . . . permeable to volatile oils and flavorings." 84 
Pliofilm, an older rival first marketed in the mid-thirties, has a rubber 
base and is particularly well adapted to packaging foods preserved in 
liquids, a relatively narrow market. Despite its superiority in this use, its 
high cost (in 1949 a thousand square inches of 120-gauge Pliofilm N2 
sold for about 1% times as much as moistureproof cellophane) made it 
"an active competitor of Cellophane only in those fringe uses bordering 
markets that need greater moistureproof protection than Cellophane pro- 
vides." 85 In 1939 Pliofilm sales were only 2 per cent of cellophane sales; 
by 1949 they had increased to only 4.4 per cent. 86 

83 DX 595, p. 1156. 

84 Ibid., pp. 1155-56. The court said of polyethylene: "Many of these deficiencies 
could be corrected through research, and were." 118 F. Supp. at 81. 

85 DX 595, p. 1153. In 1950 Goodyear developed a type satisfactory for fresh 
meats. 118 F. Supp. at 81. 

86 GX 531, p. 7101; GX 81, p. 309; DX 596, p. 1173. 



Cellulose acetate, a nonmoistureproof transparent film, is an old cello- 
phane rival. First appearing in 1931, by 1939 its sales were only 3 per 
cent of cellophane's. Ten years later they were only 3.7 per cent. 87 Its 
chief quality disadvantage is that it is not moistureproof. It compares in 


Comparison of Average Wholesale Prices of Cellophane with 
Prices of Other Flexible Packaging Materials in 1949 

Per Cent of Cello- 

Per Cent of Cello- 






per 1,000 


Sq. In. 


per Lb. 


Packaging Material 








100 gauge #517 







Cellulose Acetate 









.002"— 18" flat width 








120 gauge N2 







Aluminum Foil 








Moistureproof Cellophane 








Plain Cellophane 

300 PT 







Vegetable Parchment 








Bleached Glassine 








Bleached Greaseproof 








Plain Waxed Sulphite 

25# self-sealing 







Plain Waxed Sulphite 

25# coated opaque 







Source: Prices per thousand square inches and per pound, United States v. E. I. du Pont de Nemours if? Company, 
118 F. Supp. 41, 83 (D. Del. 1953). Robert Heller & Associates, management consultants, conducted the price survey 
for du Pont on which DX 995, the original source of these data, is based. G. W. Bricker, who personally supervised 
the survey, testified that "each of these materials is a principal standard material of that type." T. 4497. In selecting 
a particular grade Bricker relied on the advice of the Bureau of Labor Statistics, American Pulp and Paper Associa- 
tion economists, and the individual companies from which he got his data. 

quality with plain cellophane, but its 57 per cent higher price in 1949 
placed it at a serious competitive disadvantage. 

About these several films du Pont in its 1948 market analysis con- 

The principal markets for non-viscose films have been competitive with 
Cellophane only to a very minor degree up to this time. Some are used very 
little or not at all in the packaging field — others are employed principally for 

87 Comparison of total du Pont and United States production of cellophane and 
imports of selected flexible packaging materials, 1925-1949, DX 981, p. 1. 


specialty uses where Cellophane is not well adapted — none have been success- 
fully introduced into any of Cellophane's main markets due to their inherent 
shortcomings. 88 

On the superiority of cellophane as compared with other films for 
most of cellophane's uses, the experts apparently agreed. Olin Industries, 
Inc., later to become the third domestic cellophane producer, after in- 
vestigation reported: "According to du Pont, Cellophane is considered 
the only all purpose film, and any product to be truly competitive with 
Cellophane must have the following attributes: (1) low cost, (2) trans- 
parency, (3) operate with a high efficiency on mechanical equipment, 
(4) print well both as to speed and appearance." 89 Olin concluded: 

There are no films currently marketed which are potentially competitive to 
any substantial degree in Cellophane's major markets when measured by the 
above attributes necessary for wide usage. Other transparent films will find 
their place for those low volume uses which can absorb the additional cost of 
the film and which necessitate certain physical properties not possessed by 
Cellophane. 90 

Consumer decisions confirmed the judgment of the experts. In 1949 
converters used roughly fourteen times as much cellophane as all other 
packaging films. 91 

Apparently cellophane has no effective rival in another segment of the 
flexible packaging material market, the outer wrapping of packaged cig- 
arettes. Clear as plate glass, flexible, easily ripped open, moistureproof, it 
displays and protects with such perfection that except when they can't 
get it cigarette makers use no other overwrap. 92 The court recognized 
this, noting however that makers of Pliofilm, glassine, and aluminum foil 
keep trying to break into this market. They have not succeeded. 

The court to the contrary notwithstanding, the market in which cello- 
phane meets the "competition" of other wrappers is narrower than the 
market for all flexible packaging materials. Cellophane dominates the 
market for cigarette overwraps, it does not compete with kraft paper for 
general wrapping, and in its more specialized markets the nonviscose 
films do not compete with cellophane except in fringe uses. 

Food Packaging 

In 1949, 80 per cent of du Pont's cellophane sales were for packaging 
food products; here cellophane encounters its most vigorous rivalry, 

88 DX595, p. 1147. 

89 Report on "the evidence in support of entry by Olin Industries into the 
Cellophane business, based on the purchase of patent license and 'know-how' from 
du Pont," December 15, 1948, GX 566, p. 7575. 

90 Loc. cit. 

91 DX 985. This is a market analysis prepared for du Pont by Robert Heller & 

92 A shortage of cellophane in the mid-forties forced some cigarette makers to 
use other materials. Brown and Williamson Tobacco Company once experimented 
with selling Kools and Raleigh cigarettes in a one-piece foil package. 118 F. 
Supp. at 108. 


"competing" with vegetable parchment, greaseproof paper, glassine, wax 
paper, and aluminum foil. Each of these wrapping materials is a dif- 
ferentiated bundle of qualities, competing in a wide variety of uses. Users 
attach a different importance to the several qualities. Many value trans- 
parency highly, a quality in which cellophane is outstanding. Some, how- 
ever, regard transparency as a disadvantage. All are likely to rate mois- 
ture protection as important, but wax paper, aluminum foil, and some 
types of glassine are about as good as cellophane in this. Food packagers 
in selecting wrapping material no doubt consider carefully the unique 
combination of qualities represented by each of these materials. They 
resell the product they wrap and they are cost-conscious. Presumably 
they try to select the material that, quality considered, will give the 
greatest value. In determining values they must consider consumer re- 
sponse to the several materials. In any event, some buyers of packaging 
materials changed from one kind to another in trying to get their mon- 
ey's worth. Some candy makers and some bread bakers, for example, 
operating on narrow margins, in the mid-thirties switched from cello- 
phane to a less costly wrapper when their other production costs 
mounted. The court concluded from the evidence that "shifts of busi- 
ness between du Pont cellophane and other flexible packaging materials 
have been frequent, continuing and contested." 93 In no one of the more 
important uses for packaging foods did cellophane in 1949 supply as 
much as 50 per cent of the total quantity (in square inches) of wrap- 
ping materials used (see Table 2). 94 Only in the packaging of fresh prod- 
uce did cellophane sales top the list. Its percentage of total sales varied 
from 6.8 per cent for packaging bakery products to 47.2 per cent for 
fresh produce. Like du Pont's percentage of total sales of all flexible 
wrapping materials, these specific figures scarcely demonstrate that du 
Pont has monopolized the sale of flexible packaging material to food 

Such facts apparently led the court to conclude that du Pont, although 
selling about 76 per cent of the cellophane and together with Sylvania — 
whose production was geared to du Pont's — selling all of it, had not 
monopolized the market for all flexible wrapping materials. No one is 
likely to quarrel with this finding. But in an economic sense a firm may 
have a monopoly of a differentiated product, that is, it may behave like 
a monopolist and enjoy the fruits of monopoly in selling it, even though 
it meets the rivalry of substitutes. That is the economic issue here. Is 

** Ibid., p. 91. 

94 Table 2 is based on evidence which the court reproduced in the opinion. In 
less important uses not included in the court's tabulation cellophane accounted for 
the following percentages of total quantities of the selected flexible wrapping ma- 
terials used: dry beverages, 6.4 per cent; breakfast cereals, 12.6 per cent; dry fruits 
and vegetables, 63.7 per cent; frozen dairy products, 1.1 per cent; flour, meal, and 
dry baking mixes, .5 per cent; nuts, 77.3 per cent; paste goods, 97.4 per cent; paper 
products, 38 per cent; and textile products, 62.3 per cent. DX 984. 



cellophane so highly differentiated that du Pont in selling it can follow 
an independent pricing policy, that is, is the cross elasticity of demand 
for cellophane so low that du Pont, while pricing it independently, can 
enjoy a monopoly profit in its sale? Let us examine this issue. 

When du Pont first marketed cellophane, it apparently thought cello- 
phane had unique qualities and it adopted a strategy designed to prevent 
competition from any other producer, in short, to protect its monopoly. 95 
It also priced cellophane from the outset to yield monopoly revenue. Its 
long-run aim in selling cellophane was apparently that of any monopolist, 
viz., to maximize revenues. But the maximization of revenues over time 
even by a monopolist may call for a farsighted and vigorous policy in 


Comparison by Percentages of Total Quantity of Selected Flexible Packaging 

Materials, Classified by End Uses* 






Type of 






























































* Based on 1949 sales (in millions of square inches) of nineteen major converters "representing a substantial 
segment" of the converting industry, United States v. E. I. du Pont de Nemours Eff Company, 118 F. Supp. 41, 113 
(D. Del. 1953). G. W. Bricker of Robert Heller and Associates, management consultants employed by du Pont, 
testified that the above data covered two-thirds of du Pont's and Sylvania's cellophane. T. 4474. 

exploiting a product. Monopolists, although they can restrict output and 
charge relatively high prices, may not find it profitable to do so. Du Pont 
argued and the court concluded that the test of monopoly is the power 
to exclude competition and the power to raise prices. A more logical test 
is the power to exclude competition and the power to control prices. 
That a monopolist may find it profitable to lower prices, increase sales, 
and reduce costs, even though the public benefits, does not necessarily 
mean, as Robertson suggests, that he has relinquished monopoly power. 
To use monopoly power rationally is not to forego it. 

President Yerkes of the Du Pont Cellophane Company, Inc., con- 
cluded as early as 1924 that to maximize earnings du Pont should reduce 
cellophane prices. On this issue he said: 

95 If cellophane had encountered the effective competition of rival wrapping ma- 
terials, du Pont would have had nothing to gain by impeding entry. That is to say, 
if cellophane were merely one of many substitutable products among which effective 
competition prevails, the price of each would be driven down to a competitive 
(cost-remunerative) level and it should be a matter of indifference to du Pont 
whether this results from rival products or from new producers of cellophane. 


I am in favor of lowering the price. ... [I] think it will undoubtedly in- 
crease sales and widen distribution. . . . Our price I think is too high based 
purely on manufacturing cost and too high in comparison with other wrapping 
papers on the market, and while we cannot approach the price of glassine or 
other oil papers, if we make a substantial reduction we will in some cases get 
somewhere near there. 96 

Walter S. Carpenter, Jr., chairman of du Pont's board of directors, ex- 
pressed a similar idea when he testified in the Cellophane case: 

. . . the purpose of reducing our price and also improving our quality was 
to broaden our market. ... As a general philosophy I was always in favor of 
the reduction of the price as we were able to do so by the reduced costs, and I 
think that I consistently urged that on the management. 97 

The Yerkes-Carpenter philosophy apparently prevailed. The price of 
cellophane, which averaged $2,508 a pound in 1924, was reduced in every 
year until 1936, when it averaged 41.3 cents a pound. With minor inter- 
ruptions the decline continued until cellophane sold for an average price 
of 38 cents a pound in 1940. Inflation accompanying the Second World 
War reversed the trend. With few exceptions cellophane prices moved 
upward until 1950, when they averaged 49 cents a pound. 98 But despite 
the reductions moistureproof cellophane (300 MST-51, the principal 
type) sold at from two to seven times the price of 25# bleached glassine 
and from two to four and a half times the price of 30# waxed paper, 
its most important rivals. 99 

Du Pont's Independent Pricing Policy 

On its face du Pont's pricing policy was consistent with that of a 
monopolist. Other evidence supports this conclusion. Had cellophane's 
major rival wrapping materials competed with it effectively (i.e., had the 
cross elasticity of demand between cellophane and other wrappers been 
high), the prices of such wrapping materials would have moved concur- 
rently to prevent, as Chamberlin says, "incursions by one seller, through 
a price cut, upon the markets of others." 100 In fact, however, while du 
Pont was "broadening its market" by reducing cellophane prices, the 
prices of other wrappers did not follow a similar pattern. Bleached 
glassine prices were constant from 1924 until 1933 and again from 1934 
to 1938. They rose in 1939 and again in 1940. Waxed paper prices fluc- 
tuated between .5 cent and .52 cent per thousand square inches from 
1933 through 1939 and in 1940 increased to .62 cent. Vegetable parch- 

96 Memorandum of some remarks made at a meeting of die board of directors, 
Du Pont Cellophane Company, Inc., December 11, 1924, DX 337, p. 643. 

97 T. 6278-79. 

08 Table of annual average prices from 1924 to 1950, United States v. E. I. du 
Pont de Nemours & Co., 118 F. Supp. 41, 82 (D. Del. 1953). 

ou Defendant's Brief on the Facts and the Law, Appendix A (graph based on 
prices per 1,000 sq. in.), United States v. E. I. du Pont de Nemours & Co., 118 
F. Supp. 41 (D.Del. 1953). 

100 Chamberlin, Monopolistic Competition, p. 90. 


ment prices declined from 1.3 cents to 1.0 cent per thousand square 
inches between 1924 and 1928 and thereafter fluctuated between .95 
cent and 1.05 cents. Bleached greaseproof prices rose from .45 cent per 
thousand square inches in 1933 to about .55 cent in 1940. 101 But du Pont's 
cellulose acetate film dropped in price from 59.3 cents a pound in 1935 
to 53.6 cents in 1940, 102 and aluminum foil prices dropped from 2.45 cents 
to 1.65 cents per thousand square inches between 1928 and 1940. The 
prices of these rival products and of cellophane followed the same trend, 
but cellophane and cellulose acetate film sold for substantially more than 
aluminum foil; and it seems likely that the cross elasticity of demand 
between the cellulose films and aluminum foil is even less than between 
them and the other products compared. 

Under inflation wrapping material prices have increased since 1940, 
but not similarly. Average cellophane prices increased by about 20 per 
cent between 1940 and 1949, but the prices of most other wrappers in- 
creased more rapidly: vegetable parchment, about 40 per cent; bleached 
glassine, 40 per cent; cellulose acetate, 50 per cent; waxed paper, 15 per 
cent; and bleached greaseproof, 80 per cent. The only two wrappers to 
increase less in price than cellophane were Pliofilm, 13 per cent, and 
aluminum foil, 9 per cent. 103 These price patterns indicate that cellophane 
continued to decrease in price relative to most other wrapping materials. 

The above facts demonstrating cellophane's independence of other 
wrapping material prices strongly suggest that du Pont was not selling 
cellophane in an effectively competitive market. Either cellophane's rival 
products were not close enough substitutes to feel the effect of cello- 
phane price decreases (i.e., the cross elasticity of demand between cello- 
phane and these products was low) or they were already selling at cost 
and could not prevent cellophane's invasion of their markets. In either 
event they did not constitute sufficiently close substitutes to insure effec- 
tive competition. 

Although du Pont lowered its cellophane prices from time to time as 
it re-examined its demand and cost functions, at no time did it compete 
with its most popular rivals on a price basis. As H. O. Ladd, director of 
du Pont's trade analysis division, put it: 

The main competitive materials . . . against which Cellophane competes 
are waxed paper, glassine, greaseproof and vegetable parchment paper, all of 
which are lower in price than Cellophane. We do not meet this price competi- 
tion. Rather, we compete with these materials on the basis of establishing the 
value of our own as a factor in better packaging and cheaper distribution costs 
and classify as our logical markets those fields where the properties of Cello- 
phane in relationship to its price can do a better job for the user. 104 

101 DX 994- A. These price comparisons, like those for 1949 in Table 1, rest on 
data collected for du Pont by Robert Heller & Associates and are based on the 
prices of one principal standard material of each type named. 

102 GX 490, p. 6507; GX 495, p. 6665. 

103 DX994-A. 

104 GX 589, p. 7530. 


But while du Pont resorted to aggressive selling, emphasizing the su- 
periority of its product and extending its services, 105 the evidence does 
not indicate that at any time it carried quality competition so far as to 
equalize average cost and selling price. Price differences no doubt re- 
flected at the margin the customers' evaluation of differences in quality, 
but the record does not indicate that they reflected differences in cost. 
If they had, with as many firms as are selling flexible wrapping materials, 
monopoly profits would have disappeared and the market would have 
become effectively competitive. Let us turn then to du Pont's earning 


As du Pont reduced cellophane prices, output and sales expanded rap- 
idly. In 1924 du Pont produced only 361,000 pounds of cellophane and 
sold $1,307,000 worth. A decade later it produced 39,358,000 pounds 
and sold $18,818,000 worth. In 1940 when cellophane sold at 38 cents a 
pound, its all-time low, du Pont produced 81,677,000 pounds and sold 
$31,049,000 worth. 106 Such increases in output and sales had called for a 
continuous expansion in investment. In 1925 du Pont's fixed and working 
capital in producing cellophane was $2,122,000. In 1934 it was $24,008,000 
and a decade later $41,133,000 (see Table 3). 

Du Pont's production and pricing policies paid off. In 1925 it earned, 
before taxes (operating earnings), 107 $779,000 on its cellophane operating 

105 £> u p ont showed great ingenuity and aggression in developing new uses for 
cellophane and expanding old ones. R. R. Smith, assistant director of sales of du 
Pont's film department, testified that in 1934, when white bread regularly sold for 
10 cents a loaf and its profit margin was small, he and other salesmen actually created 
the specialty breads industry — new varieties of bread which could be sold at a price 
large enough to cover the higher cost of wrapping them in cellophane. T. 5704-5. 
In 1936 Smith studied the sales methods of door-to-door bakery salesmen and du 
Pont made a sales training film "which had nothing to do with packaging" (T. 5721) 
but showed the way to higher profits even when using cellophane. "The promo- 
tion was extremely successful." T. 5705. In 1951 du Pont had about 45 per cent of 
the variety bread-wrapping business. T. 5721. 

106 See comparisons of du Pont and Sylvania production and sales, 1924-1950, 
United States v. E. I. du Pont de Nemours & Co., 118 F. Supp. at 116, 123 (D. Del. 

107 Du Pont computes operating earnings for each operating division by deducting 
all of the expenses directly related to its operations from its sales. Among these 
expenses are production, selling, administration, and research expenditures conducted 
within and for the particular division. Du Pont calculates its rate of operating earn- 
ings on the basis of its working and fixed investment allocated to its cellophane 

Net cellophane earnings are calculated by allowing for federal income taxes, 
capital stock tax, franchise, state income, and foreign taxes, "B" bonus, and funda- 
mental research by the chemicals department. Federal income and other taxes con- 
stituted the great bulk of these deductions: 90 per cent as early as 1935 (GX 490, 
p. 6506) and during the Second World War practically all, when the company was 
paying large excess profits taxes. Consequently, cellophane operating earnings may 
be thought of as primarily representing earnings on total cellophane investment be- 
fore taxes, and cellophane net earnings as earnings after taxes. 

The problem of empirically determining profit rates is subject to many pitfalls. 
However, the procedure used by du Pont to determine cellophane earnings is sub- 


investment. In 1934 it earned $6,000,000 and in 1940, $12,000,000. Al- 
though its annual rate of earnings before taxes declined somewhat from 
a high of 62.4 per cent in 1928, in only two years between 1923 and 
1950 inclusive did the rate fall below 20 per cent (see Table 3). 

Du Pont's cellophane pricing policy is consistent with the economists' 
assumption that a rational monopolist aims to maximize profits. This did 
not always call for a price reduction. In 1947 du Pont earned only 19.1 
per cent before taxes and only 11.2 per cent after taxes on its cellophane 
investment 108 — the postwar low. Raising the average price of cellophane 
from 41.9 cents a pound in 1947 to 46 cents a pound in 1948 paid off. 
By May, 1948 du Pont's operative earnings had increased to 31 per cent. 
At that time its division manager announced that "if operative earnings 
of 31 per cent is [sic] considered inadequate, then an upward revision in 
prices will be necessary to improve the return." 109 He suggested a sched- 
ule of prices which would increase operative earnings to about 40 per 
cent. 110 This was not put into effect until August, 1948. Operative earn- 
ings for 1948 averaged only 27.2 per cent; but by 1949 they had in- 
creased to 35.2 per cent and by 1950 to 45.3 per cent. Operative earnings 
after taxes yielded 20 per cent on du Pont's investment in 1950. 

Du Pont's pricing policy in the postwar inflation is also consistent with 
the theory of monopolistic behavior, but the record indicates that profit 
maximization was not the sole factor affecting price decisions. The divi- 

ject to fewer criticisms than are usually encountered in profit estimates. It is true 
that earnings may be understated somewhat because of expenditures not directly 
related to cellophane manufacture and sale as noted above. On the other hand, some 
might argue that actual earnings are overstated in some years and understated in 
others because operating investment is necessarily based in part on historical rather 
than replacement costs. This error is reduced by the fact that du Pont has increased 
its capacity periodically by substantial amounts, so that of its historical costs a sub- 
stantial portion is always recent history. However, some of the most frequent and 
important shortcomings of profit estimates are not involved in our calculations; 
operating investment does not include assets capitalized in expectation of excess 
profits, nor has overcapacity broadened the investment base. Probably the most con- 
vincing argument as to the credibility of these earnings is that du Pont has no 
reason to delude itself as to what it is earning in making cellophane. The invest- 
ment base which du Pont uses to calculate its rates of operating and net earnings 
is its estimate of the actual total investment involved in its cellophane operations. 
Such an investment base is considerably larger than that used by the Federal Trade 
Commission in its study, Rates of Return (after Taxes) for 516 Identical Com- 
panies in 25 Selected Manufacturing Industries, 1940, 1941-52 (Washington, D.C., 
1954), which uses stockholders' investment as its base. If this base were used in 
calculating rates of cellophane earnings they would undoubtedly be greater for all 
years. For example, in 1935, the year before Du Pont Cellophane was consolidated 
with du Pont, the latter's equity in Du Pont Cellophane was only $9,696,000. GX 490, 
p. 6504. If this were used as a base upon which to calculate du Pont's rate of earn- 
ings in that year, instead of that actually used in Table 3, its rate of operating earn- 
ings would be about 60 per cent instead of 24.6 per cent. 

108 One reason for the relatively low earnings in 1947 was du Pont's inability to 
put its new capacity at Clinton, Iowa into production as early as predicted. DX 372. 

109 GX 591, p. 7539. 

110 Ibid., p. 7540. 



Du Pont's Operating Investment, Operating Earnings, 
and Net Earnings on Cellophane, 1925-1950* 

Rate of 

Rate of 














(per cent) 


(per cent) 


$ 2,000 


2,122 b 

$ 779 a 

36.7 b 

$ 650 b 

30.6 a 











906 d 






l,318 e 



5,099 a 

2,845 f 

55.8 f 

2,645 f 










5,431 h 

29.9 h 

5,196 h 




3,888 i 

18.0 1 

3,882 i 











24.9 k 





6,05 l k 

24.6 k 

4,934 k 



26,262 m 

7,642 m 

29.1 m 

6,119 a 

23.3 m 

























































12,241 r 











64,800 8 

17,600 a 
















Average Rate of Return 



* For definition of operating investment, operating earnings, and net earnings see footnote 107. Before 1937, 
investment and earnings figures include cellulose caps and bands, cellulose acetate, and adhesives, for some years. 
On the whole this inclusion decreases the rate of return figures for cellophane slightly, since some of these items were 
actually sold at a loss at times. The net effect is insignificant, however, since they represent such a small proportion 
of total earnings and investment — less than 5 per cent in 1935. GX 490, p. 6507. 

a Derived from relevant columns k GX 490, p. 6503 

b 1925-1928, GX 483, pp. 6409, 6410 » GX 489, p. 6493 

GX 481, p. 6375 m GX 384, p. 969 

d GX 482, p. 6396 n 1937-1947, GX 591, p. 7539 

e GX 483, p. 6418 ° GX 492, p. 6571 

t GX 484, pp. 6431, 6433 p 1938-1941, GX 495-A, p. 6716 

« GX 485, pp. 6441, 6443 <» 1942-1945, GX 499, p. 6839 

h GX 486, pp. 6453, 6455 r 1946-1947, GX 501, p. 690 

1 GX 487, pp. 6464, 6466 8 GX 577, p. 7323 
i GX 488, pp. 6479, 6481 

* GX 573 (I), p. 8. Exhibit impounded by court, cited in government's Proposed Findings of Fact, p. 48, and 
Brief for the United States, pp. 144, 145, United States v. E. I. du Pont de Nemours y Company, 118 F. Supp. 41 
(D.Del. 1953). 

Sources: The exhibits referred to are annual profit and loss statements of Du Pont Cellophane Company or 
the cellophane division of E. I. du Pont de Nemours & Company. 

sion manager in suggesting price increases called attention to other rele- 
vant factors: 

2. What effect, if any, will a price increase have on our case when it is 
heard before the Federal Judge? I have not covered this with our Legal De- 
partment but in view of the position they took last July and August, prior to 


the October increase, I am inclined to think they should be brought in for a 
discussion on this matter. 

3. The du Pont Company may get some undesirable publicity from the 
press. A price increase on Cellophane could be looked upon as added fuel to 
the present recent spurt in the inflationary spiral and add to the present pres- 
sure for an increase in wages. This question is currently a live one at several 
of our Cellophane plants. Probably it would be in order to discuss this with 
Mr. Brayman. 111 

After considering these questions du Pont executives decided on the 
price increase. 112 

Cellophane's earnings record offers persuasive if not convincing evi- 
dence that du Pont has had monopoly power in selling cellophane. A 
comparison of du Pont's earnings from cellophane with its earnings from 
rayon lends force to this conclusion. 113 Despite the dissimilarity of the 
end products, several factors justify the comparison. Cellophane and 
rayon stem from the same basic raw materials. Both are radical innova- 
tions. Both were initially manufactured under noncompetitive conditions 
and both enjoyed substantial tariff protection. The same business man- 
agement produced both products. The French Comptoir shared in the 
management of both Du Pont Cellophane and Du Pont Rayon until 1929. 
Yerkes, president of Du Pont Cellophane, was also president of Du Pont 
Rayon. Presumably du Pont in controlling business policy for both com- 
panies was actuated by similar business motives. 114 Both products have 
had several reasonably close substitutes. The production and consump- 
tion of both increased phenomenally. 115 Cellophane and rayon have been 

111 Loc. cit. Mr. Brayman was the director of du Pont's public relations depart- 

112 In considering the probable effect of a price increase on cellophane earnings, 
the division manager stated the matter as follows: "Can we sell the capacity output 
of our plants? . . . The District Managers are divided in their opinion. . . . How- 
ever, the majority of the District Managers, the Director and Assistant Director of 
Sales are of the opinion, barring a recession, the tonnage can be sold." Ibid., p. 7539. 
Although this reasoning is consistent with that of a monopolist interested in maxi- 
mizing profits, Judge Leahy cited it as evidence that du Pont did not have the 
power to raise prices arbitrarily. United States v. E. I. du Pont de Nemours & Co., 
118 F. Supp. 41, 179 (D. Del. 1953). 

113 Data are not available to compare du Pont's earnings from cellophane with the 
earnings of producers of other wrapping materials. These are without exception 
diversified firms producing a variety of products. However, the record discloses 
that in every year from 1935 through 1942 du Pont failed to cover costs in selling 
cellulose acetate film, which it sold in competition with two other concerns (GX 
490 through GX 497). 

114 The district court found: "Same individuals were the principal du Pont execu- 
tives in du Pont Rayon Co. and du Pont Cellophane Co. Same policies of im- 
proving quality, lowering cost of production, and reducing unit price to gain greater 
volume of sales were followed as to both companies"; and that du Pont's "price 
policy for rayon was the same as for cellophane." 118 F. Supp. at 86. 

115 United States consumption of rayon increased by about 320,000,000 pounds be- 
tween 1920 and 1938. Jesse W. Markham, Competition in the Rayon Industry (Cam- 
bridge, Mass., 1952), p. 230. Cellophane consumption grew by about 80,000,000 pounds 
between 1924 and 1938. DX 600, p. 1216. 



similarly characterized by rapidly developing technology, rapid reduc- 
tion in costs, and rapid decline in prices. 116 The chief difference in the 
manufacture and sale of the two products significant to the course of 
profits apparently lies in the structure of the rayon and cellophane in- 
dustries. Although rayon manufacture began in this country as a mo- 
nopoly, rival firms came into the industry promptly. American Viscose 


Investment of Principal Companies in Rayon, Investment of Du Pont in 

Rayon, Investment of Du Pont in Cellophane, and Annual Rate of Return 

Before Taxes on These Investments, 1920-1938 


Total Rayon 

Rate of 

Du Pont's 

Du Pont's 


Du Pont's 

Du Pont's 

Return of 

Rate of 

Rate of 

of Principal 




Return on 

Return on 















of dollars) 


(per cent) 

(per cent) 

(per cent) 


$ 40.7 




$ 2.9 





















$ 2.1 































































































Average Rate of Return 




Sources: Rayon investment and earnings, Federal Trade Commission, Investments, Profits, and Rates of Return 
for Selected Industries (a special report prepared for the Temporary National Economic Committee, 76th Cong., 
3d Sess.), 1941, pp. 17988, 17990, 17998. Cellophane investment and earnings based on Table 3. Comparable data 
on total rayon investment and earnings are not available beyond 1938. 

Corporation began as the sole domestic producer of rayon shortly before 
the First World War and du Pont followed in 1920. By 1930 these con- 
cerns had eighteen rivals. As late as 1949 fifteen firms occupied the field. 
Although the four largest firms in recent years have usually accounted 
for about 70 per cent of the total output and although most of the firms 
have generally followed a price leader, Markham from his painstaking 

11G Rayon prices dropped from $6.00 a pound on February 1, 1920 to $0.51 a 
pound on July 29, 1938. Federal Trade Commission, Investments, Profits, and Rates 
of Return for Selected Industries (a special report prepared for the Temporary 
National Economic Committee, 76th Cong., 3d Sess.), 1941, p. 17985. Cellophane 
prices dropped from $2.51 in 1924 to $0.42 in 1938. DX 336, p. 642. 


and exhaustive study concludes that freedom of entry and the pressure 
of substitute products have made the rayon industry workably or effec- 
tively competitive. 117 The course of both du Pont's and the industry's 
rate of earnings supports this conclusion (see Table 4). Federal Trade 
Commission data reveal that in 1920, when du Pont first produced rayon, 
American Viscose Corporation, until then the country's sole producer, 
realized 64.2 per cent on its investment. 118 Although du Pont showed a 
loss in 1921, its rate of earnings rose to 38.9 per cent by 1923. Thereafter 
its rate of earnings and those of the industry declined until by 1929 they 
had fallen to 19.0 and 18.1 per cent, respectively. When six more firms 
entered the industry in 1930, 119 average industry earnings fell to 5.0 per 
cent and du Pont suffered a loss of 0.9 per cent. During the following 
eight years du Pont averaged only 7.5 per cent on its rayon investment, 
and the industry as a whole put in a similar performance. 

In striking contrast, du Pont with only a single rival in producing cel- 
lophane (and that rival's output closely geared to du Pont's) earned less 
than 20 per cent on its cellophane investment in only one depression 
year. From the beginning of the depression in 1929 through the succeed- 
ing recovery and the 1938 recession du Pont averaged 29.6 per cent be- 
fore taxes on its cellophane investment. On its rayon investment it aver- 
aged only 6.3 per cent. 


Apparently the cellophane market does not conform to the Chamber- 
linian model in which substitutes are so close that no producer may long 
enjoy monopoly returns — a "sort of ideal" equilibrium adjustment with 
the demand curve tangent to the cost curve at some point above lowest 
average cost. It does not conform to Clark's model of workable competi- 
tion wherein rival products and potential competition reduce the slope 
of the demand curve, or to Robertson's model wherein substitutes are so 
close as to result in a competitive price. Rather, cellophane is so differ- 
entiated from other flexible wrapping materials that its cross elasticity of 
demand gives du Pont significant and continuing monopoly power. 

Du Pont has used its power with foresight and wisdom. It has ap- 
parently recognized that it could increase its earnings by decreasing its 
costs and prices, by educating its potential customers to the benefits of 
wrapping their products in cellophane, by improving machinery for 
packaging, by helping converters and packagers solve their technical 
problems. It has built a better mousetrap and taught people how to use it. 

117 Markham, op. cit., pp. 181, 206, 208. 

118 Federal Trade Commission, op. cit. supra note 116, p. 17644. In this report the 
Commission's method of estimating rates of earnings on the basis of total investment 
is apparently similar to du Pont's method of calculating its operating earnings for its 
various divisions. See note 107 supra. 

119 Markham, op. cit., p. 47. 


But du Pont has not surrendered its monopoly power. Its strategy, 
cellophane's distinctive qualities, and the course of its prices and earnings 
indicate this. Du Pont's strategy was designed to protect a monopoly in 
the sale of a product it regarded as unique, and its pricing policies re- 
flected the judgment of its executives on how best to maximize earnings. 
We think its earnings illustrate Knight's distinction between justifiable 
profits to the innovator and unjustifiable monopoly gains. They have 
been "too large" and have lasted "too long." 



Basing Point Pricing and Public 




The Supreme Court's reaffirmation (by a 4-4 tie vote without written 
opinion) of the Federal Trade Commission's decision in the Rigid Steel 
Conduit Case finally settled the legal status of basing point pricing sys- 
tems. 1 The earlier Cement Case 2 had defined any agreement to use basing 
point pricing as unlawful, and put a wide construction on what was evi- 
dence of an agreement. In the Conduit Case the Federal Trade Commis- 
sion ruled that the concurrent use of basing point pricing by numerous 
competitors was sufficient evidence of unlawful conspiracy. Further, it 
held that individual use of the system by a single firm in the knowledge 
that it was being used by competitors was likewise illegal. The Circuit 
Court, in upholding the Commission, asserted that the legal question 
presented by that case was identical with that presented by the Cement 
Case. Under no circumstances, therefore, could basing point price sys- 
tems be free of the vice of unlawful conspiracy, which make them "un- 
fair methods of competition" under the Federal Trade Commission Act. 

This decision will probably give further impetus to the discussion of 
legislation to make basing point pricing legal by amending the Clayton 
and Federal Trade Commission Acts. The discussion was initiated in the 
Eightieth Congress by the Republicans through the medium of a sub- 
committee of the Senate Committee on Interstate Commerce, under the 

* The Quarterly Journal of Economics, Vol. XLIII (1949), pp. 289-314. Reprinted 
by courtesy of the Harvard University Press and the author. 

The material in this paper was first presented to an informal discussion group 
on antitrust policy consisting of Prof. M. A. Adelman of M. I. T. and Professors 
S. S. Alexander, R. R. Bowie, D. F. Cavers, L. Gordon and Dean E. S. Mason, all 
of Harvard. I am much obliged to them for valuable suggestions and corrections. 
They are not, of course, either individually or collectively responsible for the argu- 
ment here presented or the errors it may contain. 
t Harvard University. 

1 The FTC decision is given in Rigid Steel Conduit Association, 38 F.T.C. 534 
(1944). The decision was upheld on appeal by the 7th Circuit Court of Appeals, 
Triangle Conduit and Cable Co. v. Federal Trade Commission, 168 F. 2d 175 (1948). 
The high court decision is recorded in Clayton Mark & Co. v. Federal Trade Com- 
mission, 336 U.S. 956 (1949). 

2 FTC v. Cement Institute, et al., 333 U.S. 683 (1948). 



chairmanship of Senator Capehart of Indiana (Rep.). It is being con- 
tinued in the present (Eighty-first) Congress; and a bill was introduced 
to effect such amendments (S.236) by Senator Johnson of Colorado 
(Dem.) early in January. 

The need for dispassionate examination of basing point pricing and the 
possible alternatives to it in terms of public policy is clear. The present 
essay is an attempt to do so, by no means the first in this knotty field. 3 

Before examining the tests by which public policy in this field should 
be measured, let us summarize briefly the characteristics of markets in 
which the basing point pricing system is used, and the chief features of 
the system in operation. The characteristics which follow are typical, 
but not every industry operating under the system will possess them 
all. 4 The product sold is essentially standardized, so that the output of 
one producer at a given consuming point is a perfect substitute for the 
output of another producer at that point. Therefore, in an equilibrium 
situation, the prices charged by two producers at any given point must 
be the same. The product also is low in value per unit weight, so that 
for shipments over all but very short distances, the transportation cost 
forms a substantial fraction of the delivered price. Thus spatial differen- 
tiation of the product, delivered to the consumer, forms an essential ele- 
ment of the system. Shipments do, in fact, take place over fairly long 
distances. This can arise either from economies of scale which make local 
producers serving local markets uneconomical, or from strong locational 
factors which lead to the concentration of production in a few areas, or 
both. The cement industry furnishes an example of the operation of the 
first reason; the maple flooring industry (located in the region of timber 
stands in Michigan and Wisconsin), of the second; and the steel industry, 
the classic example of basing point pricing, of the combination. The 
efficient scale of operations is large, capital investment in an economical 
plant is fairly great per unit of output, and the ratio of marginal cost to 
average cost is low for all rates of operation below "full capacity." The 

3 Some of the recent articles have been: Frank A. Fetter, "Exit Basing Point 
Pricing," and Corwin D. Edwards, "Basing Point Decisions and Business Practices," 
both in American Economic Review, December, 1948; J. M. Clark, "Law and 
Economics of Basing Points," American Economic Review, March, 1949; and Edwin 
B. George, "The Law and Economics of Basing Points" (3 articles), Dun's Review, 
September, October, and November, 1948. A book on the subject was published in 
April, 1949: Fritz Machlup, The Basing Point Syste?n (Philadelphia: Blakiston, 1949). 

4 In general, factual material on the operation of the basing point system is not 
too abundant, and nearly all that is available bears on only two industries of the many 
in which the system was used — steel and cement. All the factual material referred to 
in this paper is drawn from the experience of one or another of these industries. This 
experience is described in various places. See especially: A. R. Burns, The Decline 
of Competition, chap, vi, "Price Discrimination"; TNEC Hearings, Parts 19, 20 
and 27; TNEC Monographs 42, The Basing Point Problem, and 41, Price Discrimi- 
nation in Steel; and the Supreme Court opinion on the Cement Case, 333 U.S. 683, 
which provides a good brief summary of the operation of basing point pricing in 
the cement industry. 


rate of operation of the plants in the industry is frequently below full 
capacity, often because there are large cyclical fluctuations in demand 
for the product. Production equipment employed is specialized and long- 
lived, and thus exit from the market, either by product substitution or by 
allowing plant to "die" through non-replacement, is a difficult and pro- 
tracted process. The market demand for the product is generally in- 
elastic at and below prices which correspond to output considerably less 
than the full capacity of the industry. Finally, the market contains few 
enough sellers so that oligopolistic calculation plays an important role in 
the actions of firms in the market. This is clear enough in the steel in- 
dustry. The large number of firms in the cement industry seems to con- 
tradict this statement, but shipments of cement rarely move more than 
200 or 300 miles, and in any local market the number of competing firms 
is very small, fewer than 10 in most areas. The combination of small 
numbers and spatial differentiation makes these industries classic exam- 
ples of monopolistic competition. 

The mechanics of basing point pricing are well known, and need not 
be rehearsed here. Certain features of the operation of the system, how- 
ever, deserve the emphasis of repetition. Its outstanding feature is the 
creation of a fixed, well-defined price structure, with delivered prices of 
all sellers identical to all consumers at each specific location. The effec- 
tive operation of this system requires basing points and base prices pub- 
licly known (in the trade), and uniquely defined freight costs from 
every basing point to every possible consuming point. This last require- 
ment is usually fulfilled by the use of a common compilation of freight 
rates in the form of a freight book by all firms in the industry. In prac- 
tice, the system usually operates on the basis of all-rail freight in the 
calculation of delivered prices, although it could conceivably forego this 
restriction. Since rail freight costs must be paid, rail freight is used al- 
most to the exclusion of water and truck transport. 5 Lastly, there is 
always a significant amount of market interpenetration, defined as occur- 
ring when mills sell to customers located where mill prices plus freight 
costs from other mills which could handle their orders are lower than 
from the selling mill. This market interpenetration usually involves 
freight absorption (when there is a multiple basing point system) and 
always involves selling costs in excess of what would be spent without 
market interpenetration. The end result of market interpenetration is a 
complex structure of geographical price discrimination, determined by 
the locational pattern of mills and consumers, and the actual degree of 
market interpenetration. 

5 In steel, consumers were allowed to take delivery in their own trucks on pay- 
ment of 35 per cent of rail transportation costs. See TNEC Hearings, Part 27, 
pp. 14182 ff. See also Part 20, pp. 10830-45, and correspondence on pp. 11005-6 for 
the reaction of the Steel Corporation to attempts of a consumer to take delivery in 
his own barge at a point not the point of consumption. 


So much for the circumstances and consequences of basing point pric- 
ing. In order to examine this pricing practice from the point of view of 
public policy recommendations, it is necessary to formulate both a set of 
alternative situations which are feasible of attainment through govern- 
mental action, and a set of standards by which the desirability of the 
alternatives can be tested. There are four models which cover fairly well 
the essential features of the wide range of geographic pricing practices: 

(1) Single Basing Point Model. The use of a single basing point; or 
a single major basing point and subsidiary basing points with base price 
differentials over the primary base of the same order of magnitude as the 
costs of shipment between them. Basing point quotations are strictly 
adhered to; all freight computations are made on an all-rail basis. 

(2) Universal Freight Equalization Model. Every producing mill is 
a basing point. There is no f.o.b. mill selling, and standardized all-rail 
freights are used in calculating delivered prices. In this model, the rela- 
tionship of the base prices of the several producers can range from rigid 
price leadership, in which differentials among the various mills always 
remain the same, to complete independence in setting base prices. 

(3) Uniform F.O.B. Mill Price Model. Each seller maintains an an- 
nounced mill price which is the same to all buyers (at any given time). 
The buyer takes possession at the mill, and chooses the method of de- 
livery and pays the freight. Here again, the model does not specify the 
relationship between the mill prices of the various sellers. 

(4) F.O.B. Mill Selling vuith Price Discrimination Model. The sellers 
maintain no announced prices, but deal with each customer as best thev 
can. As in (3), customers can choose the method of transportation and 
pay the freight. When base prices are independently determined, model 
(2), universal freight equalization (if modified by allowing the customer 
the election of delivery method), amounts substantially to model (4). 

These models differ only in the specification of pricing practices; it is 
assumed throughout that the basic features of the industries using basing 
point pricing remain unchanged, except insofar as the change in pricing- 
practice in itself reacts on the structure of the industry and changes it. 
This is, of course, the assumption appropriate to an examination of the 
effects of possible changes in pricing formulae on the functioning of in- 
dustrial markets. It is assumed further that this range of pricing prac- 
tices represents possibilities which could be achieved in practice by gov- 
ernment action, an assumption for which there is some justification. The 
first model was more or less exemplified in the operation of Pittsburgh 
Plus pricing in the steel industry until the U.S. Steel Corporation aban- 
doned it under pressure from the Federal Trade Commission in 1924. A 
stage between the first and second was represented by the recent history 
of the steel and cement industries, until the Cement Institute decision. 

The third model is what would seem to be the only legal method of 
pricing in the present state of the law. Given a change in the interpreta- 


tion of the Robinson-Patman Act shown in recent Court decisions, 6 or 
its outright repeal, the fourth model might be taken as a rough predic- 
tion of how industry pricing practices could respond to the law on bas- 
ing point pricing as laid down in the Cement and Rigid Steel Conduit 

The proposed standards for evaluating the alternative pricing prac- 
tices are likewise four in number. 

( 1 ) What is the effect of the pricing system on the level and rigidity 
of prices? This involves the effects of the pricing system on the flexibil- 
ity of margins, the ease with which downward adjustments in margins 
can occur, and the effects of pricing practices on the level of costs. 

(2) Do the pricing practices facilitate or retard the adjustment of ca- 
pacity to demand in the long run? This question has two aspects — the 
problem of aggregate adjustment, and the problem of the regional and 
local balance between capacity and demand. The second aspect shades 
over into the problem of the influence of pricing practices on the general 
locational structure of industry. 

(3) What effect does the pricing system have on cyclical fluctuations 
in prices and output? Since the basic characteristics of the industries in 
question guarantee that their capacity cannot be adapted to cyclical fluc- 
tuations in demand, this question really asks how well or how badly the 
pricing system functions in helping the industry to achieve the appropri- 
ate minimum average rate of return over the cycle. 

(4) How does the pricing system affect the organization of the indus- 
try? To what extent are the number of producers, the scale of produc- 
tion, and the relative viability of large scale and "independent" produc- 
ers in the competitive struggle affected by pricing practices? 

The application of these four tests to the four models will give each 
model a set of "marks"; a comparison of these marks should give some 
guidance on the desirable aims of government action. 


In considering the relation between pricing formulae and the magni- 
tude and flexibility of margins, the basic oligopolistic character of the 
markets under consideration must again be stressed. In a situation of un- 
used capacity and inelastic demand, what prevents "cutthroat competi- 
tion" from driving price down to marginal cost? It is oligopolistic ration- 
ality — the realization by each seller that his cuts will be followed by his 
rivals, and that therefore he has little to gain by cutting, and much to lose 
by initiating a process of reducing margins. The importance of pricing 

6 Such as United States v. New York Great Atlantic and Pacific Tea Company, 
CCA 7th, 1949; and Standard Oil Company v. Federal Trade Commission, CCA 7th, 
1949. On this whole matter see the stimulating discussion by M. A. Adelman, "Inte- 
gration and Antitrust Policy," to be published in the Harvard Law Review, October, 
1949, and his article "The A & P Case: A Study in Applied Economic Theory," 
Quarterly Journal of Economics, May, 1949. 


formulae in this situation lies in the extent to which they make it easy 
for each member of the group to follow what every other member of 
the group is doing. For example (in addition to the important role it has 
in making possible an intelligent comparison of the offers of various sell- 
ers by buyers in a market with an extremely complex product structure) 
the "extra list" in steel serves a vital function in reducing the dimension- 
ality of the price structure, and thus making it feasible for the rival 
sellers to compare each others' price quotations. And it is significant that 
the extra list is explicitly agreed upon among the various producers. 7 
In an oligopolistic market, especially one such as steel in which sales are 
frequently made on contracts for large quantities running over periods 
of time, it is almost inconceivable that price reductions would be made 
if they had to be made uniformly to all comers at one fell swoop. Rather 
some kind of piecemeal reduction, involving various kinds of price 
discrimination, is to be expected; and, in fact, seems to be the rule. Any 
reduction of quoted prices is usually preceded by a period of "price 
shading," during which "secret" concessions are made to customers, con- 
cessions which differ widely as between customers. 8 In this process of 
discriminatory price reduction, various asymmetries in the market un- 
doubtedly play an important role: large customers get better prices than 
small ones; smaller producers, or financially weak producers, may be the 
first to offer concessions; producers located in relatively disadvantageous 
positions may be prompted to strive, through concessions, for more busi- 
ness in local areas where mill nets are high; and so forth. 

To the extent that a pricing system formalizes quotations, and pub- 
licizes (in the trade) the "proper" prices at which transactions should 
be made, it discourages the processes by which prices are reduced. 
Judged on this ground, the single basing point system is undoubtedly 
the worst offender among our four models. The only "proper" channel 
for price change is controlled entirely by the single producer who sets 
base prices. He is formally the price leader for the industry, and knows 
that any base price changes he makes are changes in the industry's prices, 
not only in his own. Moreover, the fact that under this system, the bas- 
ing point producer can penetrate into any market without suffering 
lowered mill nets might well make for greater caution on the part of 
other producers in making secret concessions. A universal freight equali- 
zation system may in fact function as a rigid system of price leadership, 
and be just as effective in narrowing the amplitude of price adjustments 
as a single basing point system. But the very fact that many producers 

7 See TNEC Hearings, Part 19, pp. 10557-80, testimony of Benjamin Fairless and 
other United States Steel executives, and pp. 10621-35, testimony of Eugene G. Grace, 
president of Bethlehem Steel Company. 

8 See TNEC Monograph 41, Price Discrimination in Steel, for evidence of 
discrimination between large and small purchasers of steel in a period of slack de- 
mand. Also, see Table 1 below for an indication of the range of price con- 
cessions in steel. 


can, at least potentially, set base prices independently, and the further 
fact that limits to market interpenetration, and thus to "punitive" inva- 
sions of markets, are set by the costs of freight absorption, probably 
make such a system more flexible in its operation than the single basing 
point system. 

It seems doubtful that uniform f.o.b. mill pricing would produce any 
better results from the point of view of price flexibility than a universal 
freight equalization system, and it might even produce somewhat inferior 
ones. The uniform, published mill price of each producer would cer- 
tainly make each seller's price policy highly visible to his rivals. Each 
seller would fix his attention on the market area of his rival, rather than 
on his delivered prices, as he would under any basing point scheme; but 
the rival's price behavior would be revealed just as clearly. Moreover, al- 
though each seller would be free to fix his own mill price (provided he 
wanted to exercise such freedom in an oligopolistic market) any change 
in price he made would have to be uniform to all customers. This en- 
forcement of uniformity, added to the "visibility" of his behavior to rival 
oligopolists, would do much to discourage any producer from initiating 
a price change. Finally, uniform f.o.b. mill selling would end market in- 
terpenetration. Market interpenetration in the context of a rigidly op- 
erating basing point system with strong price leadership, has no compet- 
itive virtues. However, in a situation characterized by the possibility of 
discriminatory price shading, market interpenetration, which spreads the 
rivalry among sellers over a large area of the market, instead of concen- 
trating it along the fringes where the market areas of rival sellers meet, 
undoubtedly increases the speed of price changes and probably their 
magnitude also. 

The foregoing argument shows that the fourth model, f.o.b. mill sell- 
ing with price discrimination allowed, would be the one in which the 
pricing method would offer the least assistance to rival sellers in behaving 
"rationally" in accordance with the oligopolistic character of the market, 
and would discourage the formation of a fixed pattern of price leader- 
ship. The question arises as to whether such a situation would not lead 
to "too much competition," with prices cut to marginal costs whenever 
unused capacity existed, and even efficient firms driven into bankruptcy. 9 
No categorical answer can be given to this question. It is the author's 
opinion that the danger is unlikely, for although the abolition of formula 
pricing would make it more difficult for firms to abide by the oligopo- 
listic calculations which militate against price reduction, it would not 
eliminate such calculations entirely. The fundamental oligopolistic 
character of the market appears to be a sufficient brake on really destruc- 
tive competition. 

Thus, in terms of the four models of pricing systems, the greatest gain 

9 See Professor Clark's article, op. cit. 


in the flexibility of margins is promised by a change from a single bas- 
ing point system, or a multiple basing point system with fairly rigid price 
leadership, to f.o.b. mill pricing with discrimination. A change to uni- 
form f.o.b. mill pricing seems to offer little along these lines; a little 
more may be expected from independent base pricing in a multiple bas- 
ing point system; i.e., the elimination of price leadership. 

In practice, in recent years basing point pricing systems have been of 
the multiple basing point variety, with fairly strong price leadership. The 
increase in flexibility of margins which can be expected from changes 
in the direction of more independence in pricing depend on the rigidity 
with which the systems have operated in the past. Some light is thrown 
on this point in a study by the Bureau of Labor Statistics of the prices 
actually paid for steel products by a large sample of consumers over the 
period 1939-42. Some of the results of this study are given in summary 
form in Table 1. This tabulation suggests that there is still room for a 
substantial increase in the magnitude of price reductions in periods of 
low output. In the second quarter of 1939, when output ran about half 
of rated ingot capacity, average price concessions ranged from 8 per cent 
in sheets to only 3 per cent in structural shapes. These averages con- 
cealed wide variations as between customers, some of whom paid prices 
as much as 25 per cent under published prices. But these were only a 
small proportion. A year later, with output at 72 per cent of rated ingot 
capacity, the spread of concessions was much reduced, even though aver- 
age prices increased but little. Thus, in hot rolled sheets — the product 
for which concessions were greatest — a quarter of the sales were made 
at concessions of 15 to 25 per cent on the published price in 1939. By 
the following year, the number of sales at prices 15 per cent or more 
under the published price had fallen to fewer than 5 per cent of the to- 
tal. All in all, the figures of Table 1 speak well for the efficacy of the 
basing point system as a method of price leadership in the steel industry. 

The level of prices depends on the level of costs, as well as on the 
seller's margin. To the extent that different pricing systems affect differ- 
ently the level of costs, the choice between them may influence the level 
of prices. Any pricing system which allows market interpenetration adds 
the cost of freight absorption to the other costs of the mill; it is the 
mill net yield and not the delivered price less the calculated freight which 
is the "price" from the point of view of the seller. This cost is present 
in three of the four models; only with uniform f.o.b. mill selling is market 
interpenetration ruled out entirelv. Some notion of the magnitude of the 
cost of freight absorption can be gained from a studv of steel shipments 
made for the TNEC. 10 The study covered the month of February, 1939, 

10 See TNEC Hearings, Part 27, appendix, pp. 14331-14428. The study was based 
on a sample of steel shipments of all major producers in the month of February, 1939. 
The sample covered shipments aggregating some 600,000 tons, or about 25 per cent 
of the total shipments of steel during the month. February, 1939 was a period of 
slack operations, about 50 to 55 per cent of rated ingot capacity. 



a period of slack demand in which it is likely that market interpenetra- 
tion was extensive. Freight absorption in this month ranged from about 
3 to 5 per cent of the delivered price for various products. This is a 
small cost; nor is it certain that it is all reflected in price. In an oligop- 
olistic market, part of it may be absorbed by sellers through a reduction 
in profits. 


Actual Prices Compared with Published Prices* 

Per Cent 

of Numb 

er of Sales Made at Actual 


which Were Given 


ges of 



Published Delivered Price t 

Ratio of 

Product Rate 

Actual to 

and Year (Per Cent 








(Second Quarter Ingot 









Figures) f Capacity) 








Price { 

Hot Rolled Sheets 

1939 51 









1940 72 








1942 98 






Cold Rolled Sheets 

1939 51 







1940 72 






1942 98 





Hot Rolled Strip 

1939 51 









1940 72 







1942 98 




Plates, Universal 

and Sheared 

1939 51 








1940 72 








1942 98 




Structural Shapes 

1939 51 








1940 72 








1942 98 




* The data presented in this table come from a study made by the Bureau of Labor Statistics in 1943 entitled 
Consumers' Prices of Steel Products. The study was prepared for the use of the Office of Price Administration and the 
War Production Board. It was reprinted in Iron Age, April 25, 1946, under the title "Labor Dept. Examines Con- 
sumers' Prices of Steel Products." The study was based on an examination of data furnished by 629 companies 
consuming steel products, widely distributed by region and industry. The sample included only carload lot pur- 
chasers, and excluded distributors, warehousers, and subsidiaries of steel-producing companies. The companies in 
the sample accounted for 15 per cent of steel consumption in 1940. For each company, price information was ob- 
tained only on products purchased more or less regularly (monthly, if possible) in carload lots (except for alloys') 
between 1939 and 1942, and for which customer specifications remained constant over the period. The very largest 
consumers (e.g., the major automobile producers) are excluded from the sample, and therefore the actual extent of 
concessions is probably understated somewhat. The study presents data for the third quarter, 1939 and the second 
and fourth quarters, 1941 as well as for the three quarters shown in Table 1. In addition to the products for which 
prices are shown there, the study gives figures for cold rolled strip, merchant bars, and cold finished strip. The original 
tabulations are made in items of class intervals of one per cent (ratio of actual price to delivered price) which are 
here consolidated for brevity. Note that in every case the largest concession is included. The selection of time periods 
and products given in Table 1 is, in the author's opinion, a sufficiently representative one to yield a fair picture of 
the results of the whole study. Neither for the other products or for other time periods does the pattern of actual 
prices relative to quoted prices and the utilization of capacity appear significantly different from that here presented. 

t The percentages given are for the second quarters of each of the years 1939, 1940 and 1942. 

% The published delivered price was built up for each sale from the appropriate published base price, the freight 
from the applicable basing point, and the published extras. The published delivered prices entering into the com- 
parisons of actual and published prices are those of April, 1942. But these differed if at all from published prices 
in the various quarters only by very small amounts; since extras remained constant for the period, the greatest 
change in base prices was a few per cent, and freight rates increased about 6 per cent in March, 1942. The error 
introduced by the use of 1942 prices runs in the direction of exaggerating the magnitude of the concessions in the 
second quarters of 1939 and 1940: the size of this error is very small. 


A rigid basing point system with price leadership eliminates price 
competition and channels rivalry into selling efforts. Since products sold 
under basing point pricing are characteristically standardized, the cost of 
selling effort is largely remuneration of salesmen. Such selling costs will 
be present to some extent under any pricing system in which market 
interpenetration exists and customers must choose between the products 
of rival sellers offered at the same delivered price. This will obviously be 
the case under any form of basing point system. Under f.o.b. mill selling 
with price discrimination, there will be occasions on which a seller offers 
to meet rather than shade a rival's price, and relies on a combination of 
selling efforts and customer loyalty to swing the buyer's decision. Under 
uniform f.o.b. mill selling, pure selling costs would theoretically disap- 
pear, if mills were widely separated and the freight rate structure such 
that boundaries between mills were lines rather than regions. In fact, nei- 
ther of these conditions is met. In many cases rival mills are located close 
to each other so that there are numerous consuming locations at which 
prices from two sellers would be the same. 11 Moreover, the existence of 
"blanket rates" covering long-distance shipments to and from large areas 
would produce overlapping market areas under f.o.b. mill pricing even 
if rival sellers were not located side by side. In any case, the total cost 
of the sales force cannot be considered as an "unnecessary" selling cost 
attributable to the existence of non-price competition. The salesmen per- 
form certain necessary functions in the distribution process; at a mini- 
mum, order taking, and where the product is complex, as in steel, the 
transmission to buyers of information on specifications and the suitabil- 
ity of the many grades and types of product for various uses. To the ex- 
tent that salesmen's pure "selling" functions are performed jointly with 
these other duties, no extra cost of selling exists. All in all, it is doubtful 
that changes in pricing practices will produce significant changes in sell- 
ing costs through lessening the importance of non-price relative to price 

The possibility of cost saving on the freight bill through the use of 
cheaper means of transport under other than basing point pricing, will 
be discussed below in considering the relation of pricing practices to lo- 
cation and utilization of capacity. 

Finally, the influence of flexibility of margins itself on costs should not 
be neglected. The existence of fairly strong downward pressure on mar- 
gins in periods of unused capacity, which would be expected under some 
pricing arrangements, would be a strong stimulus to energetic efforts at 
cost reduction by management. The use of the most efficient known pro- 

11 See, for example, the map of counties in Western Pennsylvania and Eastern 
Ohio, showing towns in the market areas of various steel sheet mills under an f.o.b. 
mill selling system with an assumed level of mill prices; TNEC Hearings, Part 27, 
appendix, p. 13832. This map was prepared by the United States Steel Corporation. 


duction techniques is not, in practice, something that occurs automati- 
cally. Every change in methods is painful (and even costly) in a large 
organization, and the existence of a fairly stable profit margin per unit 
tends to favor let-well-enough-alone policies. This aspect of the relation 
of prices to costs will probably be more important, in the long run, than 
the (at least conceptually) more easily measured differences in freight 
absorption and selling costs contingent on the use of differing pricing 
practices. It is another important argument in favor of giving high marks 
to the fourth model, or pricing practices resembling it. 


The second standard for the evaluation of alternative pricing systems 
is the relation between the pricing system and the adaptation of capacity 
to demand in the long run. There is no general theoretical rule on the 
responses of an oligopolistic group of suppliers to long-run changes in 
demand. The persistence over time of high margins (which formula pric- 
ing facilitates) would indicate some restriction of supply in comparison 
with what would happen in an otherwise similar competitive market. 
This is not a very helpful criterion in practice, however, and something 
more concrete is needed. The impact of the pricing system on the ad- 
justment of supply depends on the direction of change in demand over 
time; in particular, the case of an increase in demand over the long run 
must be treated separately from the case of stationary or declining de- 

If demand is stable in the long run and excess capacity (the existence 
of which is postulated as characteristic) is present only during cyclical 
troughs, capacity may be considered as adjusted to the long-run level of 
demand. The problem of cyclical adjustment which remains in this case 
will be treated in the following section. The situation in which excess 
capacity exists at cyclical peaks as well as during troughs with the long- 
run level of demand stable is similar to one in which the long-run level 
of demand is declining, in that the appropriate adjustment requires the 
elimination of some producing units. Whatever adjustment takes place 
must be brought about by losses which are large enough and continuous 
enough to drive firms into bankruptcy, or at least to threaten them with 
it. It is true, of course, that bankruptcies may not be sufficient to bring 
about an immediate adjustment of capacity to demand, since they can re- 
sult in the reorganization of old firms, or the creation of new firms in 
which the equipment of the original firm is valued at less than its repro- 
duction cost. These successor firms may survive for a while; but the ne- 
cessity of replacing old equipment will finally confront them with the 
same problems which caused their predecessors to fail. Again, the proc- 
ess of bankruptcy and reorganization might well be repeated in a new 
cycle, or even several new cycles. It must be assumed, however, that 


eventually old equipment will wear out, and potential investors and en- 
trepreneurs will learn something from the previous history of the indus- 
try, and thus reductions in capacity will ultimately occur. 

In a situation in which a reduction in productive facilities is required, 
it is clear that a working basing point system interferes with the proper 
adjustment. In the first place, to the extent that it functions as a device 
for maintaining high margins, the whole process of failure is slowed 
down. Moreover, a basing point system entails market interpenetration, 
which means that existing markets are shared among many producing 
points. The nearer a mill approaches failure, the greater would be its in- 
centive to attempt to cut into any market, no matter how distant, in 
which sales promised some mill net return over marginal cost. Thus the 
impact of a decline in demand, which in general will be greater in some 
areas than in others, will not be concentrated on the mills located in the 
areas of most rapid decline, but spread out more or less evenly over all 
producers. This will further slow down the process of failure, as com- 
pared with what would happen if the mills in areas of more rapidly fall- 
ing demand sold only in their home markets. The spreading of the im- 
pact over all producers also means that differences in financial strength 
have relatively more, and differences in productive efficiency and loca- 
tion, relatively less influence in determining which firms will leave the 
industry than they would under a pricing arrangement which did not 
permit market interpenetration. 

The hindrances to appropriate adaptation of output discussed above, 
except for the first — maintenance of high margins — arise out of market 
interpenetration, and their importance is measured by the extent of it. 
Market interpenetration is a feature of three of the four models; only 
under uniform f.o.b. mill pricing would it be absent. The extent of mar- 
ket interpenetration depends, in the other three models, more on the geo- 
graphical pattern of demand, costs, and freight rates, than on the specific 
pricing practices in which the models differ. There is one exception to 
this, however; market interpenetration probably is greatest, other things 
being equal, under a single basing point system, in respect of the rela- 
tions between the base mill (or mills) and all other mills. Thus in the 
special circumstances in which the home market of the base mill is the 
region of the greatest decline in demand, the single basing point system 
would lead to a slower reduction of capacity than a universal freight 
equalization or discriminatory f.o.b. mill system of pricing. 

Demand may be rising in the long run (instead of falling or remaining 
stationary) and the appropriate adjustment of supply to demand will re- 
quire additions to productive facilities. In this situation, excess capacity 
may be present at cyclical peaks as well as during troughs, because of 
a certain rate of building in anticipation of increasing demand, or it may 
exist only during cyclical lows. In either event, the effect of the pricing 
system on the expansion of capacity will depend on the geographic pat- 


tern of the change in demand. If it is fairly uniform from market area to 
market area, the existence of market interpenetration will not alter the 
local incidence of the stimulus to new investment. The only significant 
difference between the alternative pricing systems then which would af- 
fect the rate of investment would be the difference in profit margins ex- 
pected under them. The higher margins expected under the more rigid 
formula pricing arrangements might then act to stimulate a more rapid 
increase in capacity than would be made with more flexible and com- 
petitive pricing systems. 12 This additional stimulus, however, might be 
cancelled in large part, or even overbalanced by the greater pressure on 
the individual firm under more competitive pricing systems. 

If, as is more likely, the growth of demand is geographically uneven, 
market interpenetration becomes more important. This is most clearly 
seen in a situation in which the expansion of demand takes place entirely 
in new regions, not near existing mills, while demand in old regions re- 
mains stable or even declines. Under uniform f.o.b. mill pricing, a new 
mill would be constructed in the region of new demand whenever en- 
trepreneurs and investors foresaw enough demand in the region to main- 
tain an economically sized unit in operation at an average output rate 
which would yield a sufficient profit. Under the other three pricing sys- 
tems, characterized by market interpenetration, the entrepreneurs in the 
new region would have to anticipate penetration of their market by sell- 
ers in the old regions. Nor could they expect to retaliate by penetrating 
markets in the old regions to the same extent, since these regions would 
be crowded with suppliers, and their long standing relationships with 
local customers would be a further barrier to new sellers. Thus the build- 
ing of a new plant in the new region would wait for a higher level of 
demand under pricing systems characterized by market interpenetration 
than under a uniform f.o.b. mill system. The same kind of forces would 
be at work, although with diminished strength, in a situation in which 
there were large differences in the rates of change in demand among sev- 
eral regions without differences in the direction of the change. Thus the 
use of any pricing system which permits market interpenetration will 
lead to a slower rate of increase of capacity in response to a given 
rate of growth of demand unevenly distributed among regions, than 
would take place under uniform f.o.b. mill pricing. 

This lag in the regional adjustment of capacity to demand is not unde- 
sirable. In many situations it may provide a closer approximation to an 
"ideal" utilization of resources than would a more prompt increase of ca- 
pacity. The general economic rule for the substitution of new for old 
(but still useful) capital equipment is that such replacement is desirable 
only when average costs with the new equipment are less than marginal 
cost with the old. Under the conditions analyzed above, with demand in- 

12 This argument is essentially similar to Professor Schumpeter's arguments in 
favor of monopoly. See Capitalism, Socialism and Democracy, chap. viii. 


creasing rapidly in a new region, but remaining stable in an old region 
where there is excess capacity, it is economic to supply the new region 
from the old plants as long as the marginal cost of production in the old 
plants plus transportation to the new region is less than the average cost 
of production would be for a new plant in the new region. Under uni- 
form f.o.b. mill pricing, as argued above, new plant would be built in 
the new regions while there was still unused capacity in the old regions. 
Since marginal costs are characteristically much below average costs at 
outputs below capacity in the industries under consideration, this can 
mean that the new plants represent uneconomic additions to capital. With 
market interpenetration, the addition of new capacity will wait until 
much more of the old capacity is utilized. The reasons for not anticipat- 
ing a general lowering of prices to the level of marginal costs under f.o.b. 
mill pricing when unused capacity exists such as would bring the new 
areas within the market areas of the old mills have been examined above. 

The appropriate adaptation of capacity to demand implies an economi- 
cal geographic distribution of capacity, given the pattern of demand, as 
well as a suitable aggregate volume of productive facilities. Therefore, 
the possible impact of alternative pricing systems on the location of pro- 
ducers must be examined. And, since the location of producers of prod- 
ucts which are further fabricated may affect in turn the location of the 
fabricators, an examination of their location must also be included. In 
general, the location of producers depends on the relations between the 
locations of raw materials, labor supplies, markets, and transportation 
costs. For a pricing system to influence the location of producers, it must 
work through one of these factors. In cement, the location of the con- 
sumers is fixed independently of the location of producers; and of the 
two major raw materials, coal and limestone, the more important lime- 
stone is nearly ubiquitous. In these circumstances the use of one rather 
than another pricing system would have no locational influence on pro- 
ducers, and could not, in the nature of the case, have any on consumers. 

In steel, there is a more complex situation. Producers of steel are heav- 
ily raw-material oriented; they locate in points where the assembly costs 
of coal, ore and subsidiary raw materials are low; but markets are not 
without locational pull. 13 Markets consist in fabricators of steel; many 
of these in turn locate with reference to the source of their major raw 
material. To the extent that a pricing system makes the geographical pat- 
tern of steel prices different from the "real cost" pattern, consumers will 
locate in accordance with the geographic price structure rather than with 
the underlying real cost structure. The location of consumers may in 
turn influence the location of expansion in steel production, which fur- 
ther reacts on consumers, and so on. Thus, a cumulative process of dis- 
tortion of locational patterns can occur. This process is not infinite, since 

13 See W. Capron and W. Isard, "The Future Locational Pattern of Iron and 
Steel Production in the U.S.," Journal of Political Economy, April, 1949. 


the fabricators have to consider the costs of reaching out to their own 
markets which increase as they all concentrate at one point. This process 
is best exemplified under single basing point pricing. Under this system, 
locations at the basing point are superior to locations at non-basing point 
mills for the fabricators, other things being equal. Therefore, fabricators 
tend to concentrate near the basing point, and mills there tend to grow 
relative to mills at non-basing point locations. This is true even though 
the basing point ceases to be the lowest cost production point, as it may 
well have been at one time. Thus, under Pittsburgh Plus, the rate of ex- 
pansion of steel production in Chicago and Birmingham relative to Pitts- 
burgh 14 was probably slowed down substantially. 15 

The locational pull of base mills operates only relative to non-base 
mills. Under universal freight equalization, or any other system in which 
each mill is in effect a base, it would disappear. Nor is there any differ- 
ence between uniform f.o.b. mill pricing and the other two pricing 
models in this respect: market interpenetration in itself has no localizing 

The alternative pricing systems here discussed may have a further im- 
pact on location through their effect on transportation costs. Basing point 
systems, both single and multiple, in which prices are calculated on the 
basis of all-rail freight discourage the use of other methods of trans- 
portation by the consumer. This means that in steel, for example, a po- 
tentially advantageous site connected to a supplier by water will not have 
its transportation advantages considered by a fabricator who is choosing 
a plant location, since he cannot reap the savings arising from this loca- 
tion. The same is not true, of course, of the steel producer. To the ex- 
tent that his customers will accept shipment by water, he will ship by 
water and increase his mill net yield by the difference between rail and 
water freights, and thus he may consider the advantages of waterside 

In addition to its possible indirect effects on location of production, 
the discrimination against cheaper forms of transport embodied in bas- 
ing point pricing — models (1) and (2) — operates directly to distort the 
allocation of resources employed in transportation. Any alternative pric- 
ing system which gives the customer the option on the method of de- 
livery, without taxing particular modes of transport, will thus result in 
a more economical utilization of transportation resources. No indication 
of the magnitude of these economies can be given because of the lack 
of basic statistical data. 

It is important to note that whatever advantages of efficiency in geo- 

14 See A. R. Burns, The Decline of Competition, pp. 340-45, esp. Figure 48. 

15 The further effect of basing point pricing on the location of fabricators of 
steel producing goods to which fabrication-in-transit (f.i.t.) freight rates apply are 
not considered here. This is a problem more germane to a discussion of freight rates 
than to a general survey of basing point pricing. 


graphical allocation of resources the foregoing analysis indicates will re- 
sult from the use of one rather than another pricing system, will apply 
only to plants as yet unbuilt. Thus the benefits of a change would be 
spread out over a long future period. On the other hand, some existing 
plants might become unprofitable as a consequence of a change from 
one to another pricing system, but little can be said in general terms 
about the magnitude of this transition problem. 


The two remaining standards for grading alternative pricing arrange- 
ments are the effects of the pricing arrangements on cyclical fluctuations 
in output, and on the organization of industries using them. There is less 
to be said under these heads than under the two preceding, not because 
they are intrinsically less important, but rather because the questions 
raised under them are less amenable to general analytical answers than 
those previously discussed. 

The essence of the cyclical problem in industries of the sort under dis- 
cussion lies in the cyclical unadaptability of capacity to demand in the 
short run on the one hand, and the high income elasticity and low price 
elasticity of demand for their products on the other. This means that a 
capacity just sufficient to meet the minimum level of demand at the cy- 
clical trough would fall far short of producing what was required (at 
constant prices) in a boom. Further, the increase in prices and profits 
caused by the impact of sharply rising demand on limited capacity would 
not lead to an increase in capacity within the short period. Conversely, 
capacity sufficient to meet boom-time peaks in demand implies the ex- 
istence of much unused capacity during times of slack business, since 
price cuts would not greatly increase demand. The problem then is, 
how shall the costs of providing the peak load capacity be met. 

Ideally, a two-part price system would answer the problem — marginal 
cost pricing for output, with any deficits separately financed from tax 
revenues, and the amount of plant decided by a central authority on the 
basis of the usual welfare criteria. 16 A fairly close approximation to this 
ideal method in practice could be achieved as follows. In any industry 
under consideration, private firms would own and operate enough plant 
to produce for periods of slack demand. Additional plant would be 
owned by the government, and turned over without fee to the private 
firms for operation during periods of peak demand. Marginal cost pricing 
would be prescribed. The amount of government plant, and the amount 
to be brought into production at any time would be determined by 
the prescription of some average normal profit rule for the private firms. 
While such a scheme might be practicable from an economic point of 
view, it would involve substantial administrative problems, and also, 
probably cannot be considered practicable politically. 

16 See H. Hotelling, "The General Welfare in Relation to Taxation and to Rail- 
way and Public Utility Rates," Econometrica, July, 1938. 


Within the framework of purely private operation of industry, two 
general approaches to the problem of paying for the appropriate level of 
capacity over the cycle are possible. The first, represented by Professor 
Machlup in his recent book, 17 is that the proper pricing rule requires 
marginal cost pricing at all times; in other words, pure competitive pric- 
ing. Under this rule, capacity would be determined at a level such that 
the returns during periods of high demand would cover the deficits in- 
curred in periods of slack demand and provide for normal profits. Prices 
would fluctuate quite widely over the business cycle, since at any output 
level below full capacity, marginal cost would be much below average 
cost, while in the neighborhood of capacity, marginal cost would rise 
very sharply over a short range of output. 

While this price policy — if it were followed — might be desirable in 
certain industries using basing point pricing, it is doubtful that it is suit- 
able for steel and other basic industries producing investment goods. The 
demand for such products varies greatly over the business cycle. In gen- 
eral, demand for these products is price inelastic; this is especially true 
in periods of low and falling income, when investment is unresponsive 
to current cost changes in the face of poor expectations. An exception 
to the general rule may occur during the course of a recovery, especially 
after the most obvious and profitable investment opportunities have been 
exploited. At that time, a sharp rise in the price of investment goods 
might well react unfavorably on the volume of investment, and so 
shorten the recovery. Thus a steel industry which had capacity substan- 
tially short of that needed to meet peak demand in terms of a fixed price 
level (say, the average level over the cycle) might be an agent in cutting 
down the total level of investment and income over the cycle. This, in 
turn, would lead to further losses in the steel industry, and thus to fur- 
ther shrinkage in capacity. Even if the initial situation were one in which 
capacity was larger than needed to meet peak demands, rather than the 
reverse, it is not certain that a flexible price policy would lead to a de- 
sirable adjustment. If capacity could change only by fairly large steps — 
equivalent to the output produced by a single firm — the first reaction to 
the initial situation could lead to a position of undercapacity, and further 
adjustments would again lead away from equilibrium instead of toward 
it. In general terms, it may be stated that a flexible price policy in an 
industry of the kind under consideration will lead to an equilibrium in 
capacity only if the cyclical fluctuations in aggregate income are inde- 
pendent of the price fluctuations in the products of the industry. Where 
it is unlikely that this is the case, as in steel, which is needed for many 
investment goods and has no effective substitutes over a wide range of 
uses, it is clear that there is no strong prima facie case for a cyclically 
flexible price policy. 

The other approach to the question of cyclical price flexibility is that 

Op. ch., chap. 6, esp. pp. 210-11. 


which involves a fairly great degree of stability, with prices substan- 
tially above marginal costs in periods of low output, and perhaps below 
marginal costs at the highest peaks of demand. There is no single, defi- 
nite pricing rule which characterizes this view; it includes the range 
from the advocacy of strict full cost pricing, to the broad proposition 
that it is undesirable to allow prices to fall all the way to marginal costs 
during slumps in demand. Something near the first view is typically the 
view of businessmen in the industries under consideration; many econo- 
mists, of whom J. M. Clark is one, have expressed the second. 18 Accept- 
ance of the general proposition that complete flexibility in the sense dis- 
cussed above is not desirable, leaves unanswered the question of how 
much inflexibility is necessary, and what arrangements will produce it. 
Industry advocates of basing point pricing advance this system as a nec- 
essary barrier to undesirable price flexibility. The validity of this conten- 
tion turns on the factual issues of profitability and the adequacy of in- 
vestment. It is difficult to examine these factual issues in detail, because 
the necessary information is lacking. Nevertheless, in the important case 
of the steel industry, the level of profits (even during the thirties) and 
of investment do not suggest that the multiple basing point system just 
barely succeeded in maintaining the necessary capacity. The analytical 
argument stated above in connection with the discussion of the level of 
margins and prices applies here too; given the fundamentally oligopolis- 
tic character of the markets in question, the danger of extreme price 
competition which would drive prices down to marginal costs in periods 
of slack demand seems small. The recognition by each of the rival sellers 
in the market of the relations between his price policy and those of the 
others should suffice to restrain the downward movements of prices in 
cyclical troughs before they reach danger levels. This presumes that 
the differences in costs among the firms are not so great that a price 
which seems reasonable, or even high, to one seller, will be such as to 
drive another into liquidation. In such a situation, some redistribution of 
capacity among the firms in the market seems to be indicated. This pos- 
sibility raises broad problems of public policy on the conflict between 
considerations of efficiency in performance and considerations of num- 
bers of competitors and other aspects of market structure in the applica- 
tion of anti-monopoly policy which cannot be pursued here. 19 Nonethe- 
less, the assertion that any increase in cyclical price flexibility over that 

18 For an expression of the business view, see the now famous letter of John 
Treanor of the Cement Institute, cited in Machlup, op. cit., p. 41 n. in which the ce- 
ment industry is characterized as one which "above all cannot . . . stand free com- 
petition." It is clear from the context that price competition is meant. For an 
expression of Clark's views see the article in American Economic Review, op. cit., 
and "Basing Point Methods of Quoting Prices," Canadian Journal of Economics 
and Political Science, November, 1938. 

19 See E. S. Mason, "The Current Status of the Monopoly Problem," to be 
published in Harvard Law Review, August, 1949, for a discussion of this conflict. 


achieved under basing point pricing would lead to an undesirable exten- 
sion of concentration requires more proof than has yet been offered. 

This discussion of the relation between alternative pricing systems 
and the adjustment of capacity to cyclical fluctuations in demand can 
best be summarized by saying that there is little evidence to show that 
one model is preferable to another on this ground. While a pricing ar- 
rangement which led to complete price flexibility — defined as prices al- 
ways equal to marginal costs at the going output rate — might have un- 
desirable repercussions on the level of capacity in the industry, there is 
little to show that any of the four alternative pricing systems here ex- 
amined would lead to such an extreme degree of flexibility. Aside from 
this, whatever advantages arise from those models which encourage more 
flexible pricing have already been discussed above in the section on price 

The last test is the effect of alternative pricing practices on the organ- 
ization of industries using them. It is appropriate to examine under this 
head an argument made repeatedly by businessmen defending the basing 
point system: namely, that under it, market interpenetration allows larger 
scale production than would an f.o.b. mill pricing system. 20 This argu- 
ment is advanced with an earnestness and frequency that betokens firm 
belief in it by its proponents. Yet it is clear that a given group of plants 
will have, together, the same number of customers and the same amount 
of demand (assuming a given level of prices) under one pricing system 
as under another. Differences in the allocation of these customers among 
the several mills, arising from the presence or absence of market inter- 
penetration, will not affect the average output of the mills. Thus the 
choice of pricing system will not in itself make either for a greater aver- 
age utilization of capacity at a given scale of production or a larger aver- 
age scale of production. Nor, as was argued above, is it likely that a rigid 
basing point system will lead to a lower price level than uniform f.o.b. 
mill pricing — thus, perhaps, stimulating a greater expansion of demand, 
and allowing larger-scale production. A lowering of the price level could 
be expected from a system which allows market interpenetration but 
does not have the rigid price leadership of the basing point system, but 
this is hardly what the argument under consideration is advanced to de- 
fend. Perhaps the business belief derives its earnestness from the individ- 
ual entrepreneur's experience of the necessity of market interpenetration 
to his firm in periods of slack demand: he sees it as a means of getting 
more orders and thus increasing his production rate. Failing to appreci- 
ate the relations of his efforts in this direction to those of other entre- 
preneurs, he generalizes his experience to the whole industry. 

The single basing point system, and the multiple basing point system 

20 See for example Vol. Ill of TNEC Papers published by the United States Steel 
Corporation, The Basing Point System, especially the discussion of "local monopolies" 
which would arise under f.o.b. mill pricing. 


with strong price leadership may well promote the growth of smaller 
rivals relative to the larger price leader. These rivals are likely to be the 
first to shade prices if any shading is done, while the leader adheres to 
the formula; thus they can be more aggressive in reaching out for cus- 
tomers. Against this, for the single basing point system, must be put the 
opportunity the price leader has for penetration into the market of any 
other seller without reduction in mill net. The leader may use this op- 
portunity in such a fashion as to discourage efforts at too fast growth 
by smaller rivals. 21 Under the multiple basing point system, the cost of 
freight absorption sets limits to the ability of the leader to penetrate 
other sellers' markets, and thus his ability to use penetration as a punitive 
measure. The steel industry offers an example of the decline of the price 
leader's share of the market which was continuous over the whole his- 
tory of the firm. U.S. Steel's share of the market for steel measured in 
terms of ingot production declined from 66 per cent in 1901 to 33 per 
cent in 193 8. 22 This decline went on under both Pittsburgh Plus and mul- 
tiple basing point pricing, suggesting that whatever opportunities for 
punitive retaliation were open to the Steel Corporation under Pittsburgh 
Plus were not effectively utilized. 

Under uniform f.o.b. mill pricing, while price leadership could con- 
tinue to function, secret price cutting would not be possible, and there- 
fore, the leader and the followers would be similarly placed in this re- 
spect. Thus large firms might maintain their relative dominance better 
under uniform f.o.b. than under basing point pricing. Under f.o.b. mill 
pricing with discrimination, a firm pattern of price leadership would not 
develop; a dominant producer might or might not compete aggressively 
against smaller rivals, and thus might or might not maintain his share of 
the market. This would also be true under universal freight equalization 
with independent base price determination. 

The fact that a change in the relative shares of the market over time 
which lessens the dominance of the largest producer is facilitated by bas- 
ing point pricing (with price leadership) is not in itself necessarily de- 
sirable. To the extent that this change results from a kind of competi- 
tion in which the level of prices and margins is little affected, whatever 
gains are made in terms of transferring production to more efficient units 
are retained by these firms, and the change produces no general benefits. 
If a market in which the same number of firms share more rather than 
less equally is considered preferable on grounds of the undesirability of 
"bigness" as such, even though the change has no effect on their be- 
havior, then the kind of change the basing point system promotes is de- 
sirable. It often happens that the firm with the largest share of the mar- 

21 Professor Machlup, op. cit., makes much of this point. See especially chap. 5, 
pp. 151-68. In fact, he makes somewhat too much of it, in the author's opinion. 
22 TNEC Hearings, Part 26, appendix, p. 13853. 


ket is a long-established producer with less favorable plant sites and a 
greater proportion of old equipment than some of his smaller rivals. Then 
the kind of competition promoted by the fourth model pricing system 
or some similar arrangement is likely not only to cause a change in the 
relative shares of dominant and smaller firms in the market, similar to 
that favored by basing point pricing, but also to result in lower costs and 

A multiplant firm may find it possible to practice geographic price dis- 
crimination even under uniform f.o.b. mill pricing by what might be 
called "internal freight absorption." It could do this by shipping goods 
from a distant plant and billing the customer from a nearby one, or by 
making the actual transfer of the product from one plant to another. 
Another variation of this practice would be possible where output was 
produced in several stages; but all the final stages of production could be 
carried on at one location, while finishing was done elsewhere and the 
final product marketed from the finishing points. In either case a multi- 
plant producer could sell in a certain local market area with much less 
investment than required of a single plant producer: in the first case by 
operating only a small plant in the area and using excess capacity else- 
where to augment the supplies of the local plant; in the second, by op- 
erating only a finishing plant (or even a warehouse) locally, while the 
major part of the production took place in other areas. Such practices, 
by permitting larger producers to penetrate the markets of smaller ones 
more easily than the smaller can retaliate, might decrease the ability of 
local producers to survive. Under present law, however, these practices 
could be restrained. Either the Robinson-Patman Act, forbidding price 
discrimination not justified by cost differences which tend to injure com- 
petition; section 5 of the Federal Trade Commission Act, which forbids 
unfair methods of competition, as found by the Commission and up- 
held by the courts; or even, in some situations, section 2 of the Sherman 
Act, which makes monopolization or attempted monopolization of a 
market illegal, could be invoked against a large firm which used such de- 
vices to the detriment of local producers. With any pricing system which 
permitted market interpenetration, the local producer would be placed 
under no special disabilities in competing with his larger multiplant ri- 
vals. He could choose to retaliate against penetration of his market, and 
would have only those handicaps imposed on him by his smaller size and 
limited financial resources. 

All in all, uniform f.o.b. mill prices appear to be no more, and pos- 
sibly somewhat less, favorable to the existence and activity of independ- 
ent competitors in the market than the other pricing models considered. 
Whatever virtues are possessed in this respect by basing point systems 
with price leadership are present in a stronger form in systems which al- 
low market interpenetration but do not encourage price leadership. None 


of the systems offers any savings from changes in the scale of produc- 
tion, save as the lowering of price levels under model (4) or variants of 
it might stimulate a greater rate of growth of demand. 


The results of all four tests taken together indicate that basing point 
pricing with price leadership (that has been characteristic of it) rates 
fairly low. In the form of a single basing point system, it receives low 
marks on all the tests. A universal freight equalization system with price 
leadership rates somewhat better in respect to the adaptation of capacity 
to demand in the long run, including its effects on location, and perhaps 
in respect to the level and flexibility of price and to the viability of in- 
dependent competitors too. A change from basing point pricing to uni- 
form f.o.b. mill pricing would not be a great improvement, judged by 
the foregoing tests. In respect to price flexibility and the size of margins, 
uniform f.o.b. mill pricing would be little better than the basing point 
system. As far as it affected the ability of small producers to compete 
successfully against large ones, it might even be a little worse. Only in 
two respects does uniform f.o.b. mill pricing show clear advantages over 
basing point pricing: it would function much more effectively in facili- 
tating the reduction of capacity in the face of a long-run decline in de- 
mand, and it would eliminate the wastes incident to the uneconomical 
use of transportation arising from the discouraging of road and water 
transport, as well as that arising from cross hauling. The situation of sec- 
ularly declining demand is hardly typical for the American economy, 
and therefore the importance of the first advantage is not great. 

On the other hand, a change from the basing point system to some 
pricing system which, while allowing market interpenetration, made dif- 
ficult the formation of a rigid pattern of price leadership and still allowed 
customer choice of means of transport, could be expected to lead to 
fairly substantial improvements from a social point of view in the per- 
formance of the industries concerned. Most important would be the in- 
creased scope of price competition, with its effects in lowering costs 
and prices. Wasteful use of transportation would be avoided under this 
system as well as under f.o.b. mill pricing. The adaptation of capacity to 
changing regional patterns of demand would proceed in such a way as 
to permit substantial utilization of excess capacity in old regions before 
new capacity was added. Finally, the possibility of price as well as non- 
price competition would offer to new firms in favorable locations using 
new techniques a method of striving to displace old, well-entrenched 
producers. This development would benefit the general public as well 
as the successful competitors. 

What must be done to introduce a pricing system like that of model 
(4) in place of the now illegal basing point pricing practices? It is likely 
that businessmen in the industries affected would accept it, since it 


would permit them to continue the practice of market interpenetration 
and to maintain their old customer relations, which their testimony be- 
fore the Capehart Committee shows they wish to do. The effect of such 
a system in discouraging price leadership would not be immediately ob- 
vious, and thus would not be a reason for objection to it by the business 
community. The chief obstacle lies in the pronouncements of the law 
and of those who enforce it on price discrimination. The Robinson Pat- 
man Act, in section 1, explicitly forbids price discrimination, not jus- 
tified by cost and not made in good faith to meet competition, when it 
is likely to injure competition. The Federal Trade Commission, in its 
presentation of the Cement and Rigid Steel Conduit Cases, has shown 
itself zealous to enforce these provisions. The courts have blessed the 
FTC views. Further there has been a tendency in the FTC and the Anti- 
Trust Division of the Department of Justice to interpret "injury to com- 
petition" as injury to a competitor, as evidenced by the A. & P. and 
Standard Oil Company (of Indiana) cases. 23 

This body of law and opinion must be changed in some way if a more 
competitive pricing system is to replace basing point pricing. The es- 
sence of such a system involves discrimination, as was argued in section 
II above. For instance, a universal freight equalization system might be 
made legal by Congressional enactment, provided that the purchaser 
could, if he desires, specify the method of delivery and that he pays in 
freight figured into the delivered cost no more than the actual cost of 
delivery (though he may pay less). Such an arrangement must soon lead 
to occasional discrimination by sellers in favor of customers located on 
cheap transport routes accessible to rival sellers; and these varying dis- 
criminations will prevent the new price system from hardening in a set 
mold. The required change probably must be made by legislative action, 
since the process of changing the point of view of the courts, demanding 
as it does a prior conversion on the part of the antitrust division and the 
FTC, is always a slow one. At present, when the tide of opinion in these 
agencies seems to be running the other way, it would be even slower. 
Since there is already a demand for legislation on the subject of basing 
point pricing, it may be hoped that vigorous discussion of the issues in- 
volved will produce an atmosphere in which constructive legislative ac- 
tion is both possible and likely. 

23 United States v. New York Great Atlantic and Pacific Tea Company, CCA 7th, 
February 24, 1949 and Standard Oil Company v. Federal Trade Commission, CCA 
7th, March 11, 1949. 


The Nature and Significance of 
Price Leadership* 


That the Supreme Court's decision in the Tobacco Case 1 of 1946 at- 
taches a new significance to price leadership in oligopolistic markets 
seems beyond reasonable doubt. The Tobacco decision constitutes a re- 
versal of the stand taken by the Court in the U.S. Steel and International 
Harvester cases, where the Court ruled that the acceptance of a price 
leader by the rest of the industry did not constitute a violation of the 
Sherman Act by the price leader. 2 If we accept the full meaning of 
what the Court has really said, that parallel pricing, whether implemented 
by an agreement or not, is now illegal, pricing policies prevailing in mar- 
kets where sellers are few will henceforth be subjected to a much closer 
examination than they have been in the past. 

Accomplished students of the monopoly problem, anticipating what 
such oligopolistic market studies might be expected to reveal, have pre- 
dicted the possibility of some sweeping changes in the conduct of Amer- 
ican business enterprise. Professor Rostow, for example, sees in the Alu- 
minum and Tobacco decisions, when viewed collectively, the possible 
foundations for a new Sherman Act "which promises drastically to 
shorten and simplify antitrust trials" since they represent a triumph of 
the economic over the more cumbersome legal approach to the antitrust 
problem. 3 Professor Rostow points out specifically that such tacit par- 
allelism, as evidenced by the practice of following a price leader, now 
lies within the scope of the antitrust laws. 4 Professor Nicholls cautiously 
points out that the assumptions which he made in his recent appraisal of 

* The American Economic Review, Vol. XLI (1951), pp. 891-905. Reprinted bv 
courtesy of the publisher and the author. 
t Princeton University. 

1 American Tobacco Co., et al., v. United States, 147 F. 2d 93 (6th Cir., 1944) ; 
328 U.S. 781 (1946). 

2 United States v. United States Steel Corporation, 251 U.S. 417 (1920); and 
United States v. International Harvester Company, 274 U.S. 693 (1927). 

3 Eugene V. Rostow, "The New Sherman Act: A Positive Instrument of Prog- 
ress," University of Chicago Law Review, Vol. XIV (1947), pp. 567-600. For a 
warier appraisal of the Aluminum and Tobacco decisions, see Edward H. Levi, "The 
Antitrust Laws and Monopoly," ibid., pp. 172 ff. 

4 Rostow, ibid., p. 577. 



the Tobacco decision, 5 namely, (1) that the courts really said what he 
believed them to have said and (2) that they will carry to their logical 
conclusion the legal implications of that decision, may rest upon dubious 
grounds. Nevertheless, he concedes the possibility that such modi ope- 
randi as price leadership, the presence of which was perhaps the most 
important piece of incriminating evidence in the Tobacco case, are now 

If the legal implications of the Tobacco decision as interpreted by 
Professors Rostow, Nicholls and others be accepted, the economic con- 
sequences of price leadership and the specific conditions likely to render 
it an effective weapon against price competition in oligopolistic markets 
need to be re-examined. Because the courts have not yet faced up to the 
problem of providing appropriate remedies, the question of wherein lies 
the most fruitful remedial action should at least be raised. It is primarily 
to this task that this article is addressed. Since, however, there is always 
the danger of assigning unwarranted homogeneity to such an economic 
phenomenon, its 'significance will be appraised on the basis of (1) the 
particular types of price leadership which prevail in industrial markets 
and (2) the extent to which each type might conceivably circumvent 
forces of competition. 

Professor Stigler has distinguished between two kinds of price leader- 
ship: (1) that associated with a dominant firm and (2) that of the baro- 
metric type. 6 Since, however, one of the market conditions that the baro- 
metric firm's price is supposed to reflect is both secret and open price 
cutting, 7 it is not always possible to determine whether the barometric 
firm should be viewed as the "price leader" or as one of the first "price 
followers." Hence, for purposes of this discussion, the above otherwise 
satisfactory dichotomy will be augmented by a third type of price lead- 
ership which may be viewed either as an extreme form of the baro- 
metric type or simply as price leadership in lieu of overt collusion. 


Although most of the vast volume of economic literature on price 
practices and policies conveys the impression that price leadership is a 
logical and effective means for eliminating price competition among rival 
sellers, theoretical treatment of the topic has been cast in rather simple 
static terms and limited to three special cases. 8 

5 William H. Nicholls, "The Tobacco Case of 1946," American Economic Review, 
Vol. XXXIX (May, 1949), pp. 284-96. 

6 George J. Stigler, "The Kinky Oligopoly Demand Curve and Rigid Prices," 
Journal of Political Economy, Vol. LV (October, 1947), pp. 444-45. 

7 See Professor Stigler's illustrative case, ibid., p. 445. 

8 The number of institutional and other conditions under which the prices set 
by one firm in an industry might be used by all others is probably very large, but 
only three sets of conditions seem to make price leadership of some sort inevitable 
and at the same time identify the price leader. 


Perhaps the most familiar theoretical model of price leadership is cen- 
tered upon the dominant firm or partial monopolist. Starting from the 
assumption that an industry comprises one large producer and a number 
of smaller ones, no one of which produces a high enough percentage of 
total output to influence the price, it logically follows that the role of 
price making falls to the dominant firm. This is true because each small 
firm regards its own demand schedule as perfectly elastic at the price 
set by the dominant firm and thus behaves as though it operates under 
conditions of perfect competition. The dominant firm might set any 
price it chooses, but presumably would set one which maximizes its prof- 
its by equating its own marginal cost with its marginal revenue as derived 
from the market demand schedule and the summation of the individual 
marginal cost curves of the independent small producers. 

Professor Boulding 9 has presented two other theoretical models of 
price leadership. One relates to an industry comprising one low-cost 
high-capacity firm and one or more high-cost low-capacity firms, the 
other to an industry comprising at least two firms having identical cost 
curves but different shares in the market. In the former case, because no 
price can equate marginal cost with marginal revenue for both (or all) 
firms, a conflict in price policy inevitably arises. However, since the 
price preferred by the low-cost high-capacity firm is lower than the 
price preferred by the high-cost low-capacity firm (or firms), the low- 
cost firm can impose its price policy on the industry. In the other case, 
under assumptions described by Professor Boulding as "rather peculiar," 
that marginal cost curves for all firms are identical and that each firm's 
relative share in the market is different from that of all other firms and 
remains unchanged over the entire range of possible prices, marginal cost 
and marginal revenue are equated at a lower price for the firm having 
the smallest share in the market than for any other firm. Hence, the firm 
having the smallest share in the market at all possible prices can impose 
the price most acceptable to it on the rest of the industry. Professor 
Boulding makes no claim that the latter model is built upon sufficiently 
realistic assumptions to throw much light upon price policies generally 
but suggests that it might explain price behavior in the retail gasoline in- 

It is worthwhile to point out that in none of the above three models 
is price leadership a result of collusion; in fact, in each of the models 
price leadership is an inevitable consequence of a particular cost or de- 
mand phenomenon which precludes price collusion among sellers as a 
possible solution. Moreover, in none of the three models is the absence 
of competition attributable to the presence of a price leader. In each of 
the three cases, conditions in either the factor or product market are al- 
ready assumed to be inconsistent with the assumptions associated with 

9 Kenneth E. Boulding, Economic Analysis (rev. ed.; New York, 1948). For a 
diagrammatical presentation of the two models, see pp. 582, 586. 


highly competitive industries. Since the empirical evidence presented in 
a later section also suggests that effective price leadership, for the most 
part, is a result of monopoly rather than a cause of it, it is important 
that these two observations be borne in mind when it comes to prescrib- 
ing appropriate remedies for industries having price leaders. 


Contrary to the general belief that price leadership, because it elimi- 
nates the kink in the oligopoly demand curve, makes for a higher degree 
of price flexibility, Professor Stigler has presented evidence to show 
that "except for the number of price changes of two-firm industries 
. . . , the prices of industries with price leaders are less flexible than 
those of industries without price leaders." 10 Significant though this dis- 
covery may be as evidence of the nonexistence of kinked oligopoly de- 
mand curves, it should be pointed out that the basic conclusion reached 
by Professor Stigler applies to a particular type of oligopolistic market 
and, hence, is not conclusive evidence that price leadership, regardless 
of type, leads to less flexible prices. For example, Professor Stigler lim- 
its the industries characterized by price leadership to those in which a 
dominant firm (one that produces a minimum of 40 per cent of the total 
output of an industry and more if the second largest firm is large) is 
present. Hence, industries characterized by other types of price leader- 
ship were included among those having no price leader. Moreover, the 
average number of firms in industries classified as having a price leader 
was slightly less than one-half of the average number of firms in indus- 
tries not so classified. It is not surprising, therefore, that the former group 
shows a higher degree of price inflexibility than the latter for two rea- 

First, the rationale of price making by the dominant firm or partial 
monopolist differs but little from that employed by the pure monopolist. 
They both, presumably, have complete control over prices, but the partial 
monopolist, unlike the pure monopolist, must take account of the quan- 
tity that the competitive sector of the industry will offer at any price he 
may set. However inadequate classical theory might be in explaining the 
rigidity of monopoly prices, given the empirical evidence that monopoly 
prices are relatively inflexible, it probably follows that prices controlled 
by partial monopolists assume similar rigidities. 

Secondly, the greatest number of firms in any industry classified 
among those having a price leader was four; the average number of firms 
in such industries was three. On the other hand, one industry not classi- 
fied among those having a price leader contained as many as twelve firms 
and another contained eleven; the average number of firms in industries 
classified as having no price leader was over six. However, since many 

10 Stigler, op. cit., p. 446. 


of the excluded industries such as the rayon, newsprint, copper, gasoline, 
plate glass, window glass and plow industries possess barometric price 
leaders and a larger number of firms than those having a partial monop- 
olist, Professor Stigler's findings could also be interpreted as evidence 
that (1) prices are more flexible under barometric than dominant firm 
price leadership and (2) price flexibility increases as the number of 
firms is increased. Professor Stigler isolated and very adequately treated 
the latter relationship himself; 11 the former will be discussed more fully 

In the light of the formal theoretical construction employed to ex- 
plain the rationale of dominant firm price leadership, a fairly strong ar- 
gument can be made against even including markets where prices are set 
by a dominant firm among those containing a "price leader." Formal so- 
lutions which yield an equilibrium price in such markets preclude all pos- 
sibilities of the failure of small firms to follow the dominant firms' price 
change, and, hence, from the viewpoint of the dominant firm, increase 
the probability of their following to absolute certainty. That is to say, 
whether the dominant firm attempts to maximize profits in the short run 
by equating its own marginal cost and derived marginal revenue sched- 
ules or pursues some other price policy, so long as it produces at a rate 
of output which clears the market at its own price, the remaining firms 
in the industry have no choice but to equate their marginal costs with 
the price it sets. Essentially, therefore, the pure dominant firm market 
presents a problem of monopoly price control rather than one of price 

For purposes of public policy, to draw such a distinction between 
monopoly pricing and price leadership involves more than a mere ques- 
tion of definition. Price "leadership" in a dominant firm market is not 
simply a modus operandi designed to circumvent price competition 
among rival sellers but is instead an inevitable consequence of the in- 
dustry's structure. Hence, the only obviously effective remedy for such 
monopoly pricing is to destroy the monopoly power from which it 
springs, i.e., dissolve, if economically and politically feasible, the domi- 
nant firm. Public policy should hardly be directed toward this end, how- 
ever, before the foundations of the dominant firm's existence have been 
thoroughly examined. Nearly every major industry in the American 
economy has, in its initial stages of development, been dominated by a 
single firm — the Slater Mill in cotton textiles, the Firestone Company in 
rubber tires, Birdseye in frozen foods, the American Viscose Corpora- 
tion in rayon yarn, etc., to mention only a few. The monopoly power 
of the initial dominant firm in most industries, however, was gradually 
reduced by industrial growth and the entrance of new firms. It is not at 
all certain that public policy measures could have either hastened or im- 

Ibid., p. 444. 


proved upon the process. Where forces of competition do not eliminate 
such power, however (Professor Stigler has suggested the aluminum and 
scotch tape industries to me as possible examples), it is highly improb- 
able that a mere declaration of the illegality of price leadership by the 
courts offers itself as a sufficient or even a possible remedial measure. 
The dominant firm would simply be confronted with the dilemma of 
( 1 ) changing prices frequently and reminding the public with each price 
change that it sets the price for the industry or (2) simply varying its 
output and risk the attendant onus of price fixing. Hence, should all dom- 
inant firms accept the implications of the recent Tobacco decision at 
their face value, there would be no reason to conclude a priori whether 
prices in markets dominated by a particular firm would henceforth be 
more or less flexible, or would more closely approximate prices which 
one would expect under more competitive conditions. 


Unlike price leadership of the dominant firm type, there is no explana- 
tory hypothesis which identifies the barometric price leader. In contrast 
to the dominant firm, the barometric firm "commands adherence of ri- 
vals to his price only because, and to the extent that, his price reflects 
market conditions with tolerable promptness." 12 Hence, the reasons why 
a particular firm is the barometric firm must be found in the historical 
background of an industry and the institutional and other features which 
have shaped its development. 

It is worthwhile to note in passing that in a large number of industries 
which do not contain a partial monopolist, the price leader is frequently 
but not always the largest firm. In the newsprint industry, for example, 
International Paper, the largest producer, has led most price changes in 
markets east of the Rocky Mountains and Crown Zellerbach, the largest 
western producer, has usually announced new prices on the west coast. 
The price leadership of International, however, has sometimes been chal- 
lenged by Great Northern, another large producer. American Viscose, 
which at one time completely dominated the rayon industry, has contin- 
ued to be the accepted list-price leader although it had lost its dominant 
firm position as early as 1930. On the other hand, Phelps Dodge, only the 
third largest producer of copper in 1947, has been quite active in setting- 
copper prices since OPA controls were removed in November, 1946. 

Patently, it is not possible in every case to judge when barometric 
price leadership is monopolistic and when it is competitive in character 
without making a thorough investigation, but there are certain visible 
market features associated with competitive price leadership. For ex- 
ample, unless a particular firm has demonstrated unusual adeptness at 
adjusting prices to market forces, in the absence of conspiracy one would 

Ibid., p. 446. 


certainly expect occasional changes in the identity of the price leader. 
Moreover, unless the lines of price communication are extremely efficient, 
prices are not likely to be uniform among sellers in a specific market 
area for a short period immediately following the date the price leader 
announces a new price. A "wait and see" policy on the part of several 
sellers not only gives rise to occasional price differentials, but also sug- 
gests the absence of even tacit collusion. Furthermore, if new prices are 
communicated among buyers more rapidly than among sellers, there 
would be frequent changes in the ratios of sales (and, depending upon 
inventory policies, of production) of particular firms to the total volume 
of sales (or production) for the industry as a whole. In the rayon and 
textile industries, where each large fabricator buys yarn and cloth from 
several sellers simultaneously, this is usually the case. Buyers iron out 
price differentials among sellers by refusing to buy at old prices if the 
price leader has announced a price reduction and buy heavily at old 
prices if the price leader has announced a price increase. 

The price histories of copper and rayon yarn illustrate fairly well most 
of the outward manifestations one would expect of competitive baro- 
metric price leadership. Immediately upon the removal of OPA controls 
Kennecott Copper took the lead in advancing domestic copper prices 
from the controlled 14.375 cents per pound to the world price of 17.5 
cents per pound. 13 All other producers followed. Eight days later, on 
November 20, 1946, Phelps Dodge advanced its price to 19.5 cents per 
pound and was followed by the rest of the industry. On January 28, 
1947, American Smelting and Refining Company advanced its domestic 
price to 20.5 cents; however, other producers continued to sell at the old 
price until Phelps Dodge increased its price to 21.5 cents on March 3. 
American Smelting and Refining Company matched the new price, but 
Kennecott Copper announced a firm price policy on March 27 and stated 
that it would continue to make shipments at the old price. Large copper 
buyers announced three weeks later, however, that in their opinion "Ken- 
necott had 'reluctantly' advanced their prices to meet present levels and 
the present action to fix prices at present levels meant that Kennecott 
would be unlikely to follow any further upward price revisions from 
other sources." 14 In the latter part of June the price of copper settled at 
21.5 cents after several weeks of varying prices among sellers. Around 
the end of July, 1948, several smaller companies increased their prices 
to 23.5 cents; the larger producers did not follow immediately but with- 
drew all offerings from the market. On August 3 Phelps Dodge and 
Anaconda jointly raised their prices to 23.5 cents and Kennecott followed 
on August 11. 

Of the five major copper price changes which occurred between No- 

13 Company prices arc from various issues of the New York Times. 
™New York Times, March 29, 1947, p. 23. 


vember, 1946, and December, 1948, therefore, Phelps Dodge, a medium- 
sized producer, initiated three. Competitive factors, however, such as 
the import tariff on sales made in the United States by foreign producers 
and price movements of scrap, tin, and aluminum, probably exerted much 
more influence on copper prices during the twenty-six month period 
than did the arbitrary judgment of the firm initiating the price changes. 15 

Until 1930 American Viscose was the dominant firm in the rayon 
yarn industry. Since then the company has produced from only 30 per 
cent to 35 per cent of the total domestic output of rayon yarn but has 
first announced over 75 per cent of all list-price revisions. The price 
leader can exercise only negligible control over rayon prices, however, 
since they are largely determined by the prices of such close substitutes 
as silk, cotton, wool, nylon, orlon, and vinyon, each of which competes 
strongly with rayon in a number of market areas. Moreover, small rayon 
producers do not hesitate to sell at less than their quoted price when in- 
ventories commence to accumulate, a practice which has prompted most 
of the downward revisions announced by American Viscose. On the 
other hand, rayon list prices are seldom increased unless the industry is 
operating close to full capacity and inventories are still declining. For 
list-price movements, however, American Viscose plays the role of the 
barometric firm. 

Barometric price leadership which follows the above lines probably 
does not greatly circumvent the public interest nor is it likely that the 
Tobacco decision has brought this type of price leadership within the 
reach of the antitrust laws. The barometric firm possesses no power to 
coerce the rest of the industry into accepting its price and, in most 
such industries, it simply passes along information to the "Big Three" 
or the "Big Four" on what the rest of the industry is doing in a declin- 
ing market, and proceeds with initiating price increases in a market 
revival only so rapidly as supply and demand conditions dictate. 

For purposes of prescribing appropriate remedial action it is impor- 
tant also to differentiate between actual collusive price leadership and 
"apparent" collusive price leadership which stems more from overt sell- 
ing arrangements than from simply following price changes announced 
by a rival firm. In the steel, cement, glass container, and fertilizer in- 
dustries, what has appeared at times to be barometric price leadership 
was in fact a natural consequence of basing point and zone pricing 
systems. Under a single basing point system, if recognized and adhered 
to by all producers, giving the appearance of following a price leader 
is inevitable since the pricing policies of all sellers are unalterably geared 

15 Copper producers seem to feel that their prices are largely dependent upon the 
prices of such competing metals as aluminum and tin. Between March, 1947, and 
August, 1948, Kennecott publicly denounced further price increases since it believed 
they would induce fabricators to substitute tin and aluminum for copper. Cf. New 
York Times, May 5, 1948, p. 41, and August 3, 1948, p. 29. 


to the base mill. The same is true of a multiple basing point system if 
all base mills are owned by a single seller. Identical prices among pro- 
ducers in an industry operating under a multiple basing point system 
where the base mills are owned by different producers is not clearly a 
necessary consequence of the basing point system, but one should, on 
economic grounds, expect all prices at least to move in the same direction. 
A decrease in the base price in one area allows all producers abiding by 
this base price to further invade adjacent areas until mills in adjacent 
areas meet the price reduction; an increase in the base price in one area 
increases the demand for the commodity from mills in adjacent areas, 
thereby encouraging corresponding price increases. Hence, a sufficient 
explanation for similar price movements among producers abiding by 
a basing point system is the presence of the basing point system itself. 
The best evidence that this is so is the undisciplined pricing which oc- 
curs when the basing point system temporarily breaks down. 16 

For the most part, therefore, the barometric price leader, as defined 
by Professor Stigler and as visualized for purposes of this paper, ap- 
pears to do little more than set prices that would eventually be set by 
forces of competition. In such industries as the copper and rayon in- 
dustries, i.e., oligopolies within monopolistically competitive markets, 
these prices are largely dependent upon the prices of closely competing 
products. In more clearly delineated oligopolistic industries, particularly 
where the number of firms is fairly large, price leadership of the baro- 
metric type has seldom if ever been a sufficiently strong instrument 
alone to insure price discipline among rivals. Price leadership in the steel 
and fertilizer industries has been a subordinate feature of a basing point 
system. The glass container industry implemented price leadership by 
inaugurating a zone pricing and market sharing system. In spite of this, 
many firms were not faithful price followers. 17 In the tin can industry, 
where American Can Company has frequently been identified as the 
price leader, a recent study suggests that American's list price (com- 
puted principally from the price of tin plate) has only established the 
base line of competition for other can producers. 18 Moreover, American 
Can's influence over the price of tin cans is as much attributable to its 
quasi-monopsonistic position in the tin plate market as it is to the com- 
pany's share of the tin can market. 

From the standpoint of public policy the real problem in such markets 
as those discussed above, therefore, centers upon economic forces which 
support price leadership rather than upon price leadership per se. In in- 
dustries dominated by a strong partial monopolist, parallel pricing among 
firms stems from the monopoly power possessed by the partial monopo- 

16 Cf. Temporary National Economic Committee, Monograph No. 42, p. 3. 

17 Cf. Robert L. Bishop, "The Glass Container Industry," in The Structure of 
American Industry, edited by Walter Adams (New York, 1950), pp. 407-8. 

18 Charles H. Hession, The Tin Can Industry (privately published), p. 362. 


list and not from the tacit adoption of a price leader to circumvent price 
competition. The competitive sector of the industry often has no choice 
but to accept the partial monopolist's price. In oligopolies which form 
segments of larger monopolistically competitive industries, such as those 
which conform to the pattern of the rayon and copper industries, the 
barometric firm "leads" price changes only in the limited sense that its 
price movements are presumed by its rivals to have resulted from a 
synthesis of all the available market information. Price decreases initiated 
by firms selling closely competing products and by smaller firms within 
its own segment of the industry usually prompt downward list-price 
revisions by the barometric firm. List-price increases occur only after 
the market forces have been reversed. In most markets of an intermediate 
character the evidence indicates that price leadership has been decidedly 
a subordinate feature of a pricing policy built upon the much stronger 
foundations of trade association activity, zone pricing, basing point 
agreements, etc. 19 

A comprehensive study embracing the tacit and overt pricing ar- 
rangements among sellers in a wide variety of industries more or less 
oligopolistic in character would undoubtedly point up to more meaning- 
ful conclusions than those suggested by the above evidence. Neverthe- 
less, there is some basis for believing that the mere adoption of a price 
leader is not nearly such an effective means for eliminating price com- 
petition among the few as many economists are prone to believe. Except 
for the type of price leadership discussed below, the evidence suggests 
that the power of the price leader to preserve price discipline derives 
less from his ostensible status as the barometric firm than from the more 
overt arrangements which support it. Where such supporting arrange- 
ments are not found, the barometric firm seems to do little more than 
respond to forces of competition. If this is so, the Tobacco decision 
may have far less importance than has been attributed to it, but at the 
same time the search for remedial action in similar future cases may not 
be nearly so fruitless as is generally believed. The elimination of supports 
to effective price leadership, most of which are not particularly elusive 
targets, might very well eliminate the effectiveness of price leadership 


In industries which possess certain specific features, however, one 
would expect a priori a type of price leadership of a much different 
nature and considerably more inimical to the public interest than that 

19 An examination of recent industry studies [including those reproduced in part 
in The Structure of American Industry, op. cit., and in Walter Adams and Leland E. 
Tray wick, Readings in Economics (New York, 1948) ] reveals little evidence that 
price leadership, when not buttressed by stronger means of preserving price 
discipline, prevented price competition among oligopolists in times of market crisis. 


of the barometric type discussed above. In such industries price leader- 
ship may conceivably be so effective as to serve all the ends of a strong 
trade association or of a closely knit domestic cartel and, hence, in a 
political environment where overt collusion is illegal, may be the only 
feasible means of assuring parallel action among sellers. In view of the 
foregoing discussion, the most important market features prerequisite 
to effective price leadership of this type would seem to be as follows: 

1. Firms must be few in number and each firm must be sufficiently 
large to be compelled to reckon with the indirect as well as the direct 
effects of its own price policy. If there are several very small firms in 
the industry but no dominant firm, they, through ignoring their indirect 
influence on price, are likely to engage in promiscuous price cutting 
whenever market crises occur and, hence, at least for downward price 
adjustments, usurp the role of price leader. Moreover, such firms are not 
likely to follow the lead in upward price revisions unless they are com- 
pletely satisfied with their expected volume of sales at the new price. 

2. Entry to the industry must be severely restricted if the price set 
by the price leader is to remain close to a rationalized oligopolistic price 
for any significant length of time. If the long-run cost curve for the 
new entrant is substantially the same as those which confront entrenched 
firms, price rationalization can be only temporary since the rationalized 
price will attract new entrants which, in turn, will bid the price down. 20 
If, however, the time lag between investment decisions and actual invest- 
ment in the industry is significant, price rationalization for the duration 
of the lag may suggest itself as a profitable possibility. 

3. The "commodity" produced by the several firms need not be per- 
fectly homogeneous but each producer must view the output of all other 
firms as extremely close substitutes for his own. If this condition is not 
fulfilled, each producer is likely to view his product as distinctive in 
character and the "market" will not be characterized by a single price 
policy but by several. Examples of such individual pricing policies may 
be found in the automobile and brand-name men's clothing markets. 
Where the output of each firm is differentiated to the extent that it is 
only a moderately good substitute for the output of other firms, price 
leadership, of course, is meaningless. 

4. The elasticity of the market demand schedule for the output of 
the industry as described in (3) above must not greatly exceed unity. 
If demand for the output of the industry is elastic because the oligopoly 
is only a segment of a larger monopolistically competitive market, the 
prices of closely competing products severely limit or possibly even 
eliminate the gains to be derived from adopting a price leader. More- 
over, if demand for the output of the oligopoly is highly elastic, firms 
are not likely to adhere to the price leader's price if to do so would 

For an imperfect example, see discussion of cigarette industry, infra. 


result in substantially less than capacity operations, since each firm could 
still stimulate its own sales considerably by lowering its price, even 
though all other firms met the new price. The price history of the do- 
mestic rayon industry and the postwar price history of the copper in- 
dustry furnish particularly good evidence of the validity of this point. 
Whenever declining silk and cotton prices have commenced to reduce 
the volume of rayon sales at existing list prices, rayon producers, if the 
price leader has not already reduced his price, have sold at less than list 
price in order to move accumulating inventories and to maintain opera- 
tions at near-capacity output. Similarly, copper producers appear to fol- 
low the price leader only if they believe his new price is in line with 
prevailing scrap, aluminum, and tin prices. 

5. Individual-firm cost curves must be sufficiently similar so that some 
particular price allows all firms to operate at a satisfactory rate of output. 
If, for example, the industry is composed of several high-cost low- 
capacity firms and several low-cost high-capacity firms, the resulting 
conflict in price and output policies cannot be resolved by adopting a 
price leader so long as all firms remain in the industry. 21 Low-cost firms 
will not accept the price leadership of high-cost firms since there is a 
better option in the form of a lower price and a higher rate of output 
open to them. They can therefore force the high-cost sector of the in- 
dustry to adopt the lower price but, if the differences in costs between 
high-cost and low-cost firms are significant, high-cost firms will not re- 
cover full costs and will gradually be eliminated from the industry. 
Hence, the conflict will have been resolved and the condition that all 
producers be confronted with reasonably similar cost curves will then 
be fulfilled. 

It might be argued that the foregoing conditions are fully as neces- 
sary for any form of effective parallel action, such as price maintenance 
agreements, strong trade associations, or even unimplemented oligopo- 
listic rationalization, as they are for effective price leadership. Such an 
argument, of course, would be entirely valid for, it will be recalled, the 
type of price leadership being examined is but one of a number of 
possible forms of conscious parallelism, all of which presumably stem 
from a common source, namely, the identity between the long-run in- 
terests of each individual firm and those of the industry as a whole. 

Moreover, conditions other than those discussed above bear signifi- 
cantly upon the likelihood of effective price leadership ever arising and 
maintaining price discipline in an industry. Among those that first come 
to mind are the extent of tariff protection, the rate of technological 
change, the stability of demand, and the aggressiveness of management. 
An examination of the available price histories of industries in which 
the number of sellers is not large indicates, however, that price leader- 

21 For the theoretical analysis relevant to an industry containing several firms but 
only one low-cost high-capacity firm, see Boulding, supra, fn. 9. 


ship is most likely to serve the ends of a collusive agreement when the 
above five conditions are fulfilled. Or, stated another way, effective price 
discipline seems to have been rarely achieved by the tacit means of price 
leadership alone when one or several of these conditions did not exist. 


Had the Department of Justice diligently searched the American econ- 
omy for an industry which most nearly contained all the conditions pre- 
requisite to effective price leadership, it could hardly have found a better 
example than the cigarette industry. The entrenched position of the "Big 
Three" brand-names had made entry to the cigarette industry exceed- 
ingly difficult. Moreover, parallel action in the leaf tobacco market had 
insured fairly comparable if not equal cost conditions among the three 
large cigarette producers; and, although each of them viewed the output 
of the other two as such perfect substitutes for his own that none would 
risk a retail price differential, demand for their output collectively, at 
least in the short run, was inelastic. Furthermore, in 1929 the Big Three 
controlled over 90 per cent of the domestic cigarette market and, with 
Lorillard, they controlled 98 per cent. Hence, for all practical purposes, 
the number of cigarette producers was very small. Also, the large ciga- 
rette producers had had ample opportunity as well as compelling reasons 
for working out a modus operandi which would identify their individual 
interests with those of the Big Three collectively. In substance, counsel 
for Liggett and Myers probably described the attitude of all the large 
producers of cigarettes when he stated, ". . . in making price decisions 
the management of Liggett and Myers has acted in response to a long 
experience of non-identical prices as well as identical prices." 22 

In spite of such ideal conditions for securing parallel action by adopt- 
ing a price leader, however, the Big Three soon discovered that even 
their market was subject to economic forces that put an upper limit on 
exploitation. The long-run demand for their collective output was elastic, 
hence complete exploitation of the cigarette market was limited to a 
short-time period. With low tobacco prices and high cigarette prices 
in the latter half of 1931 and 1932, competitive forces began to assert 
their influence. Whereas the 10-cent brands had been virtually unknown 
(accounting for only 1.5 per cent of all cigarettes sold) in the first half 
of 1931, output of small independents began to increase rapidly after the 
price increase led by Reynolds in June, 1931. By December, 1932, they 
accounted for 22 per cent of total cigarette sales. In the meantime, the 
sales of Reynolds, American, Liggett and Myers and Lorillard had been 
drastically reduced. By February, 1933, their vulnerability to competition 
had become sufficiently evident to the Big Three to induce reductions in 

-American Tobacco Company v. United States, 147 F. 2d 93 (6th Cir., 1944), 
Liggett and Myers' Brief, p. 264; ibid., Reynolds' Brief, p. 390; and American To- 
bacco's Brief, pp. 94-95. 


popular brand cigarette prices to the lowest level since 1918. Hence, sim- 
ple price leadership, even under such ideal conditions as those afforded by 
the cigarette industry, had failed to preserve the rationalized oligopoly 
(or monopoly) market solution. 

In the light of their alleged strategy after 1933, perhaps no one was 
more aware than the Big Three themselves of the long-run ineffective- 
ness of price leadership when not implemented by other safeguards from 
competition. Although price leadership continued to play an important 
role in cigarette pricing, its effectiveness after 1933 was largely depend- 
ent upon the successful effort of the Big Three to manipulate the leaf 
tobacco market. 

If unimplemented price leadership proved to be an exploitative weapon 
of limited effectiveness in the cigarette industry, and its usefulness con- 
fined to a time period scarcely exceeding several years, it is highly im- 
probable that tacit parallel pricing in oligopolistic markets offers itself 
per se as either a fruitful or fertile field for antitrust investigation. Hence 
the Tobacco decision, particularly when viewed against a background 
in which appropriate remedial action is conspicuously absent, is not 
likely to have far-reaching consequences. The appropriate question be- 
fore economists, the business community, and the courts alike, there- 
fore, is not how far tacit parallel pricing in oligopolistic markets can 
proceed before it becomes illegal, but rather what implementing devices 
and market conditions make price leadership both possible and effective. 
In most oligopolistic industries where the record of pricing techniques 
is fairly complete, there are good reasons for suspecting that price lead- 
ership is essentially a shadow of more insidious pricing devices and trade 
restraints. When the devices which buttress price leadership have been 
destroyed, price leadership as an exploitative practice may well have 
been emasculated. 

In view of the extraordinary conditions prerequisite to the more 
effective type of price leadership, it is not likely that the Tobacco deci- 
sion, as a legal precedent, can or will measurably influence the behavior 
of prices in markets where sellers are few in number; nor will it greatly 
broaden the scope of the antitrust laws. Along these lines, the recent 
basing point and similar future decisions would appear to be a much more 
profitable line of approach to monopoly problems posed by industries 
comprising relatively few sellers. 


Price and Production Policies of 
Large-Scale Enterprise* 


The current emphasis on price policy, as against price, as a proper ob- 
ject of study represents recent economic reflection on the significance 
of expectations, uncertainties, market control, and the position of price 
as one among many selling terms. Policy implies some degree of control 
over the course of events and, at the same time, the use of judgment 
as to the probable consequences of alternative lines of action. In perfect 
markets, whether monopolistic or competitive, price is hardly a matter 
of judgment and where there is no judgment there is no policy. The 
area of price policy, then, embraces the deliberative action of buyers 
and sellers able to influence price; that is to say, it covers practically the 
whole field of industrial prices. 

The preoccupation with policy questions certainly indicates a trend 
towards an inclusion in price analysis of an increasing number of institu- 
tional considerations. Pursued to its Hamiltonian end 1 it implies not only 
an examination of the facts peculiar to each industrial market situation, 
but also a study of the conditions peculiar to each sale or purchase in- 
cluding what Messrs. Ford and Firestone dreamed in the night preced- 
ing the morning of their big tire deal. Particular circumstances may, in- 
deed, justify so minute an investigation. It is submitted, however, that 
useful work in the field of industrial price policies requires a frame of 
reference of much greater generality. To the construction of such a 
frame of reference, which must take the form, I think, of a classification 
of market structures, recent theoretical work makes a useful contribu- 
tion. It is, however, merely a starting point. 

A firm may have a price policy by reason of the existence of rivals of 
whose action it must take account, of the desirability of considering the 

* The American Economic Review, Vol. XXIX Suppl. (1939), pp. 61-74. Re- 
printed by courtesy of the publisher and the author. 

The author is indebted to the Shaw Fund and to the Harvard Committee on 
Research in the Social Sciences for financial assistance in his work in this field. 
t Harvard University. 

1 Cf. Walter Hamilton, Price mid Price Policies, Sections I and IX. 



effect of present upon future price, of the possibility of its price in one 
market affecting its price in another, of the possibility of competing in 
other ways than by price, and for many other reasons. All these situa- 
tions involve some degree of market control on the part of a seller or 
buyer. A position of market control, while a necessary, is not, however, 
a sufficient condition for price policy. In addition a seller or buyer must 
customarily conduct his operations by means of a quoted price. A dealer 
on an organized produce exchange may conduct transactions of sufficient 
magnitude to influence the market price. Yet if he buys and sells "at the 
market" it serves no useful purpose to attribute to him a price policy. I 
limit the meaning of this term, then, to buyers and sellers who enjoy 
some degree of market control and who carry on their purchases and 
sales through the medium of a quoted price. 2 Practically speaking this 
includes all selling transactions outside of agriculture and the organized 
produce and securities markets. 


The size of a firm influences its competitive policies in a number of 
ways. In the first place the scale of its purchases and sales relative to the 
total volume of transactions in the firm's market is one indication of the 
extent of its market control. Taken in conjunction with other data it may 
throw a good deal of light on price and production policies. Certain 
authorities, on the other hand, brush aside figures on the relative size 
of firms as irrelevant and emphasize the decisive importance of the 
elasticity of the firm's demand curve. 3 It would no doubt be extremely 
convenient if economists knew the shape of individual demand and cost 
curves and could proceed forthwith, by comparisons of price and mar- 
ginal cost, to conclusions regarding the existing degree of monopoly 
power. The extent to which the monopoly theorists, however, refrain 
from an empirical application of their formulae is rather striking. 4 The 

2 R. F. Kahn ("The Problem of Duopoly," Economic Journal, March, 1937, p. 4) 
distinguishes the following "extreme cases": 

"(a) At one extreme we have the case where, in spite of a change in a competitor's 
price, firms' prices remain constant automatically until they are altered as a result of 
deliberation or experiment. . . . 

"(b) At the other extreme is the case where it is the volume of sales that auto- 
matically remains constant until some other decision is arrived at." 

The first case is practically significant and embraces the whole range of industrial 
market situations in which sellers act through price quotations. The second case, 
however, is unrealistic. If price varies from moment to moment with changes in 
market conditions it is more than probable that sales (for an individual seller) will 
vary also. 

3 Cf. A. P. Lerner, "The Concept of Monopoly and the Measurement of Mo- 
nopoly Power," Review of Economic Studies, June, 1934. 

4 Some theorists, pursuing their analysis on a high plane, refer to their work as 
"tool making" rather than "tool using." A "toolmaker," however, who constructs 
tools which no "tool user" can use is making a contribution of limited significance. 
Some knowledge of the use of tools is probably indispensable to their effective 


alternative, if more pedestrian, route follows the direction of ascertain- 
able facts and makes use only of empirically applicable concepts. 5 One 
such set of facts embraces the data relevant to concentration. 

Secondly, the absolute size of a firm, as measured by assets, employees, 
or volume of sales, is also relevant to price and production policies. The 
scale of operations may affect the number and character of the factors 
that are taken into account in the determination of policies; it may also 
affect the way the firm reacts to given market situations. Selling practices 
at the disposal of the large firm may be beyond the reach of its smaller 
competitors. Large oil firms characteristically brand their gasoline and 
differentiate it from the product of competitors by extensive advertising 
campaigns. Small firms may, by reason of their size, be forced to sell an 
unbranded product at a lower price. In a society in which size is popu- 
larly considered a menace, the large firm must consider carefully the 
probable reception of its price and production policies by public opinion 
and political agencies. There is some evidence that the United States 
Steel Corporation for a considerable period of time viewed with favor 
its dwindling share in the national market and, through its price policies, 
"held an umbrella" over the heads of its growing competitors. Recent 
aggressive price tactics on the part of this company may indicate that it 
no longer regards such policies as politically necessary. If the market be 
considered to embrace all the factors external to the firm which habitually 
influence its competitive policies, there can be no doubt that the size of 
the firm affects the scope and structure of the market. 

The size of a firm likewise influences its reaction to given market situa- 
tions. Economic analysis exhibits a disposition to treat the firm as a "profit 
maximizing" agency the action of which in the market is independent of 
its internal organization. The growth of corporate bureaucracies (with 
the consequent institutionalization of management decisions), the separa- 
tion of ownership from control, and the growing influence of labor or- 
ganization on policy making are all factors "internal to the firm" which 
may and do affect its reaction to market situations. One of the questions 
raised by these considerations is the meaning and importance of ad- 
ministered prices. To this question I shall return in a subsequent section. 


Current consideration of price policy is apt to take either one of two 
quite different directions. One approach, associated with the theory of 
oligopoly and monopolistic competition, starts with various elements of 
the market structure of the individual firm and derives therefrom con- 
clusions regarding the price and production policy of this firm. The 
other begins with an examination of the behavior of prices and through 

5 I should be far from denying, however, the value of theoretical speculation, 
even of a very abstract sort, in helping to ask the right questions of the data and 
in indicating the irrelevance of much factual material. 


correlating various measures of price behavior with other measurable 
economic variables works back towards differences in the structure of 
markets in an attempt to explain the observed differences in price be- 

The practical utility of the analytical method has been to focus atten- 
tion on rivals' reactions as considerations in the determination of price 
and production policies and on the importance of non-price forms of 
competition. Whether the further elaboration of techniques of analysis 
yielding results of illusory exactness are useful is doubtful. The broad 
justification of this type of analysis must be that it provides a pattern of 
thought useful in separating data which are relevant from those which 
are irrelevant to the explanation of price and production policies. Cer- 
tainly numbers of buyers and sellers in the market and the possibility of 
product differentiation are relevant. No one would deny, furthermore, 
that the position and shape of individual demand and cost curves would 
be relevant if ascertainable. 6 In the absence of such data, however, a 
realistic analysis of price and production policies may be unable to make 
much use of the constructions of recent monopoly theory. 

A number of more specific strictures on the utility of this theory for 
price analysis may be offered. The static equilibrium assumptions im- 
plicit in this analysis rule out most of the considerations which are im- 
portant for price policy. These considerations are in the main connected 
with industrial growth and decay and with the business cycle. The ob- 
jection is not that monopoly theory is incompatible with an analysis that 
takes these considerations into account but that its constructions are ir- 
relevant to the real problems. If we seek to build further on the existing 
foundation, the only part of that foundation which is likely to be found 
usable is composed of the ascertainable facts of numbers of sellers (and 
buyers) and product differentiation. 7 

6 However true it may be that businessmen have a roughly accurate notion of 
the shape of the demand curve with which they are confronted, at least within a 
limited range, it seems extremely unlikely that economists will be able by independent 
investigations to ascertain this shape except by the roughest sort of deduction from 
other data. In certain favorable cases demand curves for a product may be drawn 
statistically; for other products we are able to surmise that the demand curve is 
elastic or inelastic. Cf. J. M. Cassels, A Study of Fluid Milk Prices, p. 41: "Certain 
fairly definite conclusions about the character of the consumer's demand for milk 
can be drawn from a general common sense analysis of the factors involved." By 
taking into consideration such factors as numbers of sellers, product standardization, 
and others one can, in some cases, proceed from a rough knowledge of the shape of 
the product demand curve to a rougher guess at the shape of the demand curve for 
an individual seller. 

7 Pigou would apparently deny that products for which substitution is not perfect 
can be in the same market. He defines the market as a "common nodal point" at 
which different units of an identical good are "available for purchase and sale" en 
route from the sellers' works to the buyers' home. Such a definition permits of a 
classification of markets only on the basis of numbers of buyers and sellers. (A. C. 
Pigou, The Economics of Stationary States, p. 78.) If different products are admitted 
to the market then the problem becomes one of (a) defining the group of products 


Data on numbers, furthermore, tell us little regarding price and pro- 
duction policies unless there is further specification of market structure. 
Elaborate speculation on the probable behavior of A on the assumption 
that B will act in a certain way, seems particularly fruitless. It recalls 
Morgenstern's discussion of the dilemma of Dr. Moriarty when con- 
fronted with the alternative courses open to Sherlock Holmes. 8 It should 
be a function of market analysis so to particularize as to reduce the area 
of necessary speculation to a minimum. The theory of oligopoly has 
been aptly described as a ticket of admission to institutional economics. 
It is to be regretted that more theorists have not availed themselves of 
this privilege. If they had there would certainly be less of a disposition 
in the literature on the decline of competition to assume that in all 
markets dominated by a few sellers are to be found the same or similar 
patterns of price policy. 

The statistical approach to price policy starts with an examination of 
price behavior and then proceeds to correlate various measures of price 
change with changes in other economic variables. Despite the recognized 
defects in price data, recent work along this line has made clear charac- 
teristic differences in the various groups of prices and has raised problems 
of admitted importance. Typically, however, this work exhibits certain 
weaknesses. First, it has been insufficiently recognized that, proceeding 
from the standard products of the raw material markets to the differenti- 
ated products of the highly fabricated goods markets, price as an index 
of the terms on which buyers acquire or sellers dispose of commodities 
tends to lose significance. The introduction of various forms of non- 
price competition and a proliferation of selling terms emphasize the ne- 
cessity to take these considerations into account both in an analysis of 
sellers' price policy and in determining changes in the position of buying 
groups. 9 Second, the measures of price behavior customarily employed 
are frequently much too general to serve the purposes to which they are 
put. This is conspicuously true of commonly used measures of price 

which are in the same market; (b) defining the geographical area within which 
buyers and sellers are in competition. It is on the basis of numbers and product 
differentiation that Machlup constructs his classification of markets, without, how- 
ever, dealing with the question of how the geographical and product limits of the 
market are to be defined. Cf. Fritz Machlup, "Monopoly and Competition," American 
Economic Review, September, 1937. 

8 Wirtschaftsprognose, p. 98. 

9 In this connection it is necessary to distinguish between two quite different 
attacks on the validity of existing price data. It may be objected that the quoted 
price is inaccurate because it is not the price at which sales actually take place; e.g., 
the B.L.S. quotations on sulphuric acid were, during the depression, highly inaccurate 
since they represented unimportant sales to small purchasers while the bulk of the 
sales during this period were made at much lower prices to large industrial users. On 
the other hand it may be objected that the quoted price is inaccurate because it is 
merely one among many conditions of sale. Insofar as the other conditions of sale 
cannot be legitimately reduced to price terms — and they usually cannot — this is not 
an objection to the validity of the price as such. 


sensitivity. An all-purpose measure of price sensitivity or flexibility is 
subject to as many and as serious objections as an all-purpose index 
number. Price is a function of many variables and price sensitivity or 
flexibility acquires significance mainly as a relationship between price 
change and change in some one or more of these variables. 10 Prices may 
be sensitive to changes in inventories, demand, costs, and in other prices, 
or to changes in some of these variables and not in others. Third, the 
attempt to correlate measures of price behavior with other data such as 
industrial concentration and product durability on an economy-wide 
basis is apt to include irrelevant and exclude relevant determinants of 
price policy. It seems probable that empirical work will achieve better 
results by a more intensive examination of specific market situations. In 
selected industrial markets a study of the relation between changes in 
costs, inventories, sales, production improvements and other variables, 
and the magnitude and timing of price change, may considerably in- 
crease our knowledge of price and production policies. 

These strictures on current methods of interpreting price and produc- 
tion policies do not imply that analytical and statistical techniques are 
useless. On the contrary any classification of market structures designed 
to illuminate patterns of competitive policy must make use of them. 


It follows from what has been said that an adequate analysis of price 
and production policies requires consideration of (a) the influence of the 
organization of a firm on the character of the firm's reaction to given 
market situations; and of (b) elements of market structure which include 
many more things than numbers and product differentiation. It goes 
without saying that a realistic treatment of these questions necessitates 
the use of analytical tools which are amenable to empirical application. 
The problem, as I see it, is to reduce the voluminous data concerning 
industrial organization to some sort of order through a classification of 
market structures. Differences in market structure are ultimately expli- 
cable in terms of technological factors. The economic problem, however, 
is to explain, through an examination of the structure of markets and the 
organization of firms, differences in competitive practices including 
price, production, and investment policies. 

A consideration of the relation of the organization of firms to price 
and production policies raises at the outset the question of administered 
prices. As currently used this is neither a clearcut nor a useful concept. 

10 Some of these relationships are, of course, much more important than others. 
I should be prepared to admit, with Lerner, Kalecki, and Dunlop that — at least for 
commodities for which price is the only significant selling term — the relationship 
between price and marginal cost is peculiarly significant both for analysis and policy. 
Cf. A. P. Lerner, op. cit.; M. Kalecki, "The Determinants of Distribution of National 
Income," Econometrica, April, 1938; J. T. Dunlop, "Price Rigidity and Degree of 
Monopoly," a manuscript to be published shortly. 


In one sense it appears to relate to the methods by which a price is deter- 
mined; in another sense to the way the price behaves. A price may be 
determined by administrative action; i.e., it may be quoted by a seller 
rather than determined by the higgling of buyers and sellers in an or- 
ganized market. At the same time it may behave in much the same way 
as prices in organized markets. A manufacturer's price of cotton print 
cloth may be taken as an example. Furthermore the attempt to contrast 
administered prices with market prices obscures the fact that all prices 
are market prices in the sense that market considerations influence their 

There is, nevertheless, an important kernel of truth concealed in this 
usage, to wit, the fact that firms are not, regardless of what economic 
theory may suppose, undifferentiated profit-maximizing agencies which 
react to given market situations in ways which are independent of their 
organization. The large corporation is a complex administrative unit in 
which control frequently bears a very attentuated relationship to owners' 
interests, in which management is increasingly professionalized, in which 
the character of labor organization may influence price and production 
decisions, and in which at best a considerable area of important price 
decision must be routinized and delegated to subordinates. The result is 
that management, in the determination of price and production policies, 
is subjected not only to market pressures but influenced in its action by 
considerations internal to the firm. 

The United States Steel Corporation, considering all the quantity and 
quality variations involved in adapting forty or fifty basic products to 
the specifications of its customers, is faced with the problem of setting 
some fifty thousand prices. While market considerations may, at one 
time or another, influence the relationships between these prices, it is 
impossible to make independent decisions respecting prices with every 
change in the market situation even if such action were thought desirable. 
The result is that pricing on individual orders is delegated to price clerks 
armed with an elaborate book of extras and such specific directions as 
to its use as may be thought desirable. The International Harvester Com- 
pany, in servicing its agricultural implements, manufactures and stocks 
some two hundred and fifty thousand separate parts. In pricing these 
parts considerations relevant to the organization and administration of 
the firm are probably at least as important as considerations relevant to 
the market situation. 

The locus and character of control within the firm may likewise be 
relevant to basic price and production policies. The familiar contrast be- 
tween a financial type of control primarily concerned with the conserva- 
tion of assets and control by entrepreneurial types mainly concerned 
with expanding output and the firm's share in the market is doubtless too 
easy. Conservation of assets may necessitate an expansion of output and, 
after all, bankers called in to rehabilitate a declining firm have been 
known to advise price reduction as a remedy. Furthermore in those 


cases in which the supposed influence of entrepreneurial and financial 
attitudes are sharply contrasted, e.g., automobiles and steel products, the 
differences can probably be more adequately explained by market con- 
ditions than by considerations relevant to the internal organization of 
firms. Nevertheless it is true to say that organizations make men, as 
well as the reverse, and in the making of men policies are also made. 
During the recent flurry of price cutting in the steel industry the presi- 
dent of the Steel Workers' Organization Committee announced that if 
price cutting continued organized labor might be forced to take action 
since "price cutting always leads to wage cutting." The character of 
control and the action of control in the determination of policy — in- 
cluding price and production policies — is influenced not only by the 
pressure of labor but by many pressures arising from group relationships 
within the firm. These relationships, furthermore, tend to influence the 
kind and caliber of men who are called to management positions in a 
concern. 11 

Economists have been singularly loath to investigate these semipoliti- 
cal relationships within large-scale enterprise which influence business 
policy. Where business policies are recognized as running counter to 
what would seem to be rational action in the market the disposition has 
been to interpret them in terms of individual personalities. It was said 
that Firestone and Ford were sports in the sense of deviations from the 
norm of entrepreneurial rationality. Or again that what every industry 
dominated by a few firms needed was a Ford or Firestone by which it 
was implied that economic rationality in such situations would lead to 
production restriction and price maintenance but that these policies were 
prevented, with advantage to the public, by the anomalous behavior of 
such entrepreneurs. No doubt Messrs. Ford and Firestone set the impress 
of their personalities on the policies of their respective industries, but the 
larger problem for economists to consider is the impress of large-scale 
business organizations on the character and functioning of the manage- 
ment groups that are called to control positions. 12 

11 Cf. J. A. Schumpeter, "Der Unternehmer in der Volkswirtschaft von heute," 
Struktur Wandlungen der Deutschen Volkswirtschaft, I: 303. In a "trustified" econ- 
omy the performance of entrepreneurial functions is subject to a "mechanization 
and bureaucratization" of decision (Willensbildung) . The type of business leader 
associated with large-scale enterprise tends to resemble the successful political figure, 
"a good minister, or bureauchief." "The groups and interests who select the leader 
tend to agree on a compromise candidate — not always the man of highest ability. 
Even when the object is to find the 'best man' he may turn out to be not one who 
can run the concern but a man adept at manipulating public opinion and handling 
public relations." 

12 The effect of the development of what may be called a professional manage- 
ment point of view on corporate policies is a question too frequently neglected. 
As an expression of that point of view, cf. Owen D. Young quoted in J. C. Sears, 
The New Place of the Stockholder: 

"To whom do I owe my obligations? 

"My conception of it is this: that there are three groups of people who have 
an interest in that institution [General Electric Company 1. One is the group of 


When we proceed from a consideration of the effect of the organiza- 
tion of a firm on its reaction to market situations to a consideration of the 
elements of market structure and their relation to price and production 
policies we are immediately confronted with the necessity of making 
clear the meaning of market and market structure. A preoccupation 
with logical elegance might lead us to define a market, with Pigou, as 
a nodal point at which a product, whose units are perfect substitutes 
for each other, are available for purchase and sale. Unfortunately, such 
a definition would effectively relegate all the important and interesting 
problems to the area of intermarket relationships. An alternative would 
be to conceive of a market as an area in geographic and product space 
bounded, in Joan Robinson's phrase, by a gap in the chain of substi- 
tutes. Within such an area, however, assuming that it could be defined, 
the position of individual sellers and buyers may be very different with 
respect to the influences affecting business policy. These and other con- 
siderations suggest that, at least in the industrial area, the market, and 
market structure, must be defined with reference to the position of a 
single seller or buyer. The structure of a seller's market, then, includes all 
those considerations which he takes into account in determining his busi- 
ness policies and practices. His market includes all buyers and sellers, of 
whatever product, whose action he considers to influence his volume of 

The classification of market structures on the seller's side consists, 
then, in grouping together those firms, in whatever industry, which 
operate under the same or similar objective conditions. 13 Among these 
conditions are the economic characteristics of the product: is it a pro- 
ducers or consumers good, is it durable or non-durable, is the product of 
an individual seller differentiated with respect to the products of other 
sellers in the same market or is it standardized? Another group of con- 
ditions relate to the cost and production characteristics of the firm's 
operation. The ratio of overhead to variable costs at given volumes of 
output and for given variations in volume of output, the flexibility of 
costs, locational factors, and the existence of joint costs are all important. 
A third class of considerations has to do with the numbers and relative 

fifty-odd thousand people who have put their capital in the company, namely its 
stockholders. Another is a group of well towards one hundred thousand people who 
are putting their labor and their lives into the business of the company. The third 
group is of customers and the general public. . . . 

"One no longer feels the obligation to take from labor for the benefit of capital, 
nor to take from the public for the benefit of both, but rather to administer wisely 
and fairly in the interest of all." 

13 The author is at present engaged, with his colleague, Professor D. H. Wallace, 
in working out a classification of industrial market situations. 


sizes of buyers and sellers of whose action our given seller has to take 
account and with the relative ease of entry for new firms. Among the 
demand conditions which are empirically determinable may be men- 
tioned the trend of sales, seasonal and cyclical fluctuations in sales, and, 
roughly, the knowledge possessed by buyers with respect to the quality 
and characteristics of the product. Differences in distribution channels 
provide another set of conditions of great importance for the policies 
and practices of a firm. The accurate specification and measurement of 
these and other market conditions with respect to an individual firm ad- 
mittedly presents great, but not insuperable, difficulties. Properly used 
the available data should permit of an illuminating grouping of firms into 
classes exhibiting roughly the same type of market conditions. Under 
similar market conditions may not firms be expected to pursue similar 
policies and practices? A careful study of the empirically determinable 
differences in market structure may go far in explaining observable dif- 
ferences in policy and practice. 

It may be objected that most of what are here called market conditions 
are already taken into account in traditional value and price analysis in 
much neater fashion. That is, at least in part, true in the sense that tradi- 
tional analysis purports to focus the results of many policy-determining 
considerations in the form of demand and cost curves which are, for dif- 
ferent time periods and under certain qualifications, single-valued func- 
tions of output. We can admit that if cost and demand curves for short, 
long, and intermediate periods were discoverable, rather than assumed, a 
large part of what is called business policy could be explained without 
resorting to so crude a device as a classification of market structures. It 
is, however, precisely because theoretical techniques of price analysis 
have been constructed without regard to their empirical applicability that 
such a classification is necessary as a first and primarily important step 
towards an understanding of business policies and practices. 

Enough has been said to suggest that the size of firms is only one 
among many factors influencing price and production policies. It requires 
no more than a cursory examination to perceive that large firms con- 
fronted with different market situations pursue different policies and 
practices. In the automobile industry the existence of large firms and a 
relatively small number of sellers was not incompatible with steadily 
falling prices which pushed the use of the product into lower and lower 
income classes until well into the 1920's. When large returns from price 
reductions seemed no longer possible, automobile manufacturers turned 
their attention to accelerating the replacement demand for new cars by 
yearly changes in the design and structure of their product. By and large 
it probably continues to be true that a strong tradition exists in the in- 
dustry to the effect that a substantial price reduction or an improvement 
of the product should be made in each year's model. The shift in em- 
phasis from price to product competition may well have been the result 


rather of a change in the economic age of the industry than a change 
in the size or number of sellers. Although the price and production poli- 
cies of the automobile indusd y are frequently contrasted with policies in 
the steel industry, may we not expect the former to approximate the 
latter as demand for motor cars becomes almost entirely a replacement 
demand and as product improvement takes increasingly the form of mere 
design or gadget changes? The economic age of an industry exerts an 
important influence on the policies and practices of firms in the industry. 
There is a widespread conviction among businessmen that aggressive 
price competition is an effective policy only during the period of an 
expanding market and that with a relatively stable or declining demand 
some type of controlled competition is in the interests of all sellers in 
the market. Controlled price competition is not a policy limited to large 
firms or markets in which sellers are few though, of course, numbers 
may be so large as to make effective control difficult. 

In the steel industry price and production policies differ markedly as 
between products undergoing substantially the same fabricating process 
and sold by substantially the same firms. A striking example of a diver- 
gence in price policies over the cycle is suggested by the behavior of 
the prices of automobile body steel as compared with the prices of gal- 
vanized steel sheets in the period between 1929 and 1937. Both of these 
products are made in the same kind of mill and the technological process 
is very similar. One, however, is sold to a few large buyers and the other 
to many small buyers. Automobile sheet prices declined more sharply 
from the beginning of the depression and at the bottom were 38.5 per 
cent below the 1929 level while galvanized sheet prices were only 28.7 
per cent below 1929. On the rise since 1933, the price of automobile 
body steel went up more slowly than that of galvanized steel sheets. 
During 1937, the price of the latter exceeded its 1929 level by 6.2 per 
cent while the price of the former was still 12.8 per cent under its 1929 

In the rubber tire market four large firms sell around 75 per cent of 
the total volume and there are in the whole industry no more than 
twenty-eight firms. Yet price and production policies would seem to be 
quite different than in other markets, e.g., cigarettes, in which firms are 
large and the number of sellers are few. Consideration of the structure 
of the tire market appears to indicate that the character of distribution 
channels exerts a decisive influence on price policies. In the market for 
tires as equipment on new cars the sellers are confronted with buyers 
each large enough to undertake tire production himself if dissatisfied 
with the price. In the replacement market a number of different distribu- 
tive channels induces a discount structure which facilitates price cutting 
on the slightest provocation. While the personality of Firestone, plus the 
fact that his firm is admittedly a low-cost producer, has no doubt been 
an important factor, it seems probable that if Firestone, like God in 


another context, had not existed the structure of the tire market would 
have created him. 

Another type of market in which the large firm has typically followed 
a policy of aggressive price competition is to be found in the field of 
distribution. Here the price we are concerned with is the spread be- 
tween manufacturers' and retail prices. Forty or fifty years ago by all 
accounts the distribution patterns for most consumers goods sold at re- 
tail was highly standardized, with full-function wholesalers and retailers 
operating under a relatively inflexible markup system. The growth of 
chain stores and other types of mass distributors has probably contributed 
not only toward a lowering of the manufacturer-retail margin but toward 
making it more flexible over the cycle. This influence is likely to con- 
tinue — unless checked by recent and prospective legislation — as long as 
mass distributors can acquire by aggressive price tactics an increased 
share of the available sales. 

These examples seem to indicate that the price policies of large firms 
are apt to be influenced by the stage of economic development of the in- 
dustry in which they operate, by the size of buying units, the character 
of distribution channels, and the possibility of obtaining an increased 
share of total sales of a group of products. There are, of course, many 
other elements of market structure which affect business policies and 
practices. In consequence it seems doubtful whether any useful generali- 
zations can be made regarding the price and production policies of large- 
scale enterprise without further specification as to the market situations 
which confront such firms. 

It may possibly be true that a rough inverse correlation might be dem- 
onstrated between concentration as measured by percentage of volume of 
output of a product produced by a given small number of firms and, 
for example, some measure of amplitude of wholesale price change over 
a business cycle. Would such a correlation, if demonstrated, reveal an 
important fact regarding short-run price policy of large-scale enterprise? 
I think not. In the first place such a correlation would be heavily biased 
by agricultural products all of which exhibit low concentration and high 
amplitude of price change. Everyone admits that the structure of agricul- 
tural products markets is at once atomistically competitive and incapable 
of realization in industry. The principal problem we are concerned with 
is whether within the range of fabricated products there is a marked rela- 
tion between size of firms and the type of price policy which is followed. 
In the second place the available price data for fabricated products is 
inadequate for such a comparison for two reasons. For many products 
there is a marked discrepancy between B.L.S. prices and average net 
realization per unit of sale reported in other sources, 14 a discrepancy 

14 Bureau of Mines and Bureau of Census figures. In part this discrepancy and 
the changes in discrepancy are the result of nonreported price cuts, in part of 
changes in the complex structure of a commodity price which are not adequately 


which varies considerably over the cycle. Furthermore, it is probably 
true to say that, in general, the more highly fabricated the product the 
less important is price as a comprehensive indicator of the terms of pur- 
chase and sale. That is to say, a study of price behavior would have to be 
supplemented by an examination of changes in product and selling terms. 

As another example of the difficulty of establishing a relationship be- 
tween size of firms and price policy through an examination of price be- 
havior, consider the recent history of construction materials prices. 
Thirty-five B.L.S. prices of construction materials were higher in Janu- 
ary, 1938, than in 1929 or 1926; twenty-one prices, on the other hand, 
were lower at this date than the wholesale price index. An examination 
of these prices fails to indicate any well-marked influence of size of firm. 
Among the high-priced products were structural steel, wire nails, cast 
iron pipe, and terne plate, all produced in industries in which a small 
number of firms produce a large percentage of output. On the other 
hand the high-priced products included cypress lumber, shingles, yellow 
pine and maple flooring, and common building brick produced in in- 
dustries in which the typical firm is small. Among the low-priced prod- 
ucts were wallboard, glass, sewer pipe, and a number of porcelain prod- 
ucts fabricated by large firms in industries with high concentration, and, 
on the other hand, a number of products typically produced by small 
concerns. It is difficult in this instance to discover any pattern of price 
behavior which would throw light on the relationship between size of 
firm and price policy. 

The relative size of a selling unit, to recapitulate, is one element — 
doubtless a very important one — in the structure of a firm's market. As 
such it exerts an influence on the policies and practices of the firm. But 
firms of given size, relative to the extent of their markets, will follow 
very different price and production policies in different market situa- 
tions. Differences in the character of price response to given changes in 
the cost or demand conditions facing a firm or group of firms are to be 
attributed both to differences in the internal organization of the firm and 
to differences in the structure of the market in which the firm, or group, 
is placed. An analysis of the relation between organizational and market 
differences and the character of price response is the central problem of 
price analysis. The relation of size to price policy is merely one part of 
the problem which, taken out of its setting, is not very amenable to 
fruitful discussion. 


In conclusion a few remarks may be offered on the relation of price 
analysis to public policy. A consideration of the consequences of dif- 

represented in the reported price; e.g., the B.L.S. price for men's shirts is the 
relatively stable price of a high-grade trade-marked product. The sale of this shirt 
fell off markedly during the depression in favor of lower-priced and frequently un- 
branded shirts. 


ferent types of price response to changes in costs and demand for the 
functioning of the economy is the prerequisite to effective public action 
in the price area. These consequences can be usefully divided into two 

1. The effect of differences in price responses on the distribution of 
economic resources among different uses. This is the traditional monop- 
oly problem. A monopoly position is supposed to lead to restriction of 
output and of investment in the monopolized area below that which is 
desirable and attainable with a greater degree of competition. A whole 
range of problems, therefore, centers around the effect of price policies 
and price relationships on the distribution of economic resources as be- 
tween various uses. 

2. The effect of differences in price response on continuity in the use 
of resources already invested or available in different uses. This is pri- 
marily a business cycle problem. It is frequently maintained that certain 
types of price response to changes in costs and demand conditions are 
more favorable to continuous employment than others. The second 
group of problems, therefore, turns around the effects of different types 
of price policy and behavior on the continuity of employment of eco- 
nomic resources. 

The argument, for both groups of problems, runs from differences in 
market structure to differences in price response, and from differences 
in price response to the consequences of these differences for the func- 
tioning of the economy. Proposals for public action, therefore, must 
consider, first, what types of price behavior and price policy are most 
conducive to an effective use of resources, and, second, within what 
limits appropriate public action is likely to be able to influence price 

Although a good deal has been written both on the effect of restrictive 
policies on the distribution of resources and on the effect of price policies 
on fluctuations in employment and output, very little has been done to 
formulate tests of undesirable price behavior applicable to public action. 
Specifically, what sort of tests are indicative of the existence of a price 
sufficiently high to restrict output and investment below desirable levels? 
What types of price behavior in industrial markets would be likely over 
the cycle to promote a fuller use of economic resources? 

Without attempting to answer these questions attention may be called 
to three issues of immediate importance in the price field facing econo- 
mists interested in public policy. 

First, is it desirable that during periods of business upturn and down- 
turn prices respond readily (in ways that can be roughly specified) to 
changes in costs, sales, or other variables of determinable magnitudes? 
If not for all commodities, for what groups of commodities should prices 
be flexible? 

Second, should certain types of price behavior, the use of price for- 


mulae, or particular price policies be accepted as prima facie evidence of 
violation of the antitrust acts? 

Third, is price competition ever sufficiently ruinous to justify public 
action? What are the tests of ruinous competition and what type of pub- 
lic action is appropriate? 

Insofar as the price and production policies of large-scale enterprise 
provide a proper field for public action, these are critical questions. 



Uncertainty, Evolution, and 
Economic Theory* 


A modification of economic analysis to incorporate incomplete in- 
formation and uncertain foresight as axioms is suggested here. This ap- 
proach dispenses with "profit maximization"; and it does not rely on 
the predictable, individual behavior that is usually assumed, as a first ap- 
proximation, in standard textbook treatments. Despite these changes, the 
analytical concepts usually associated with such behavior are retained be- 
cause they are not dependent upon such motivation or foresight. The 
suggested approach embodies the principles of biological evolution and 
natural selection by interpreting the economic system as an adoptive 
mechanism which chooses among exploratory actions generated by the 
adaptive pursuit of "success" or "profits." The resulting analysis is ap- 
plicable to actions usually regarded as aberrations from standard eco- 
nomic behavior as well as to behavior covered by the customary analysis. 
This wider applicability and the removal of the unrealistic postulates of 
accurate anticipations and fixed states of knowledge have provided moti- 
vation for the study. 

The exposition is ordered as follows: First, to clear the ground, a brief 
statement is given of a generally ignored aspect of "profit maximization," 
that is, where foresight is uncertain, "profit maximization" is meaningless 
as a guide to specifiable action. The constructive development then be- 
gins with an introduction of the element of environmental adoption by 
the economic system of a posteriori most appropriate action according 
to the criterion of "realized positive profits." This is illustrated in an ex- 
treme, random-behavior model without any individual rationality, fore- 
sight, or motivation whatsoever. Even in this extreme type of model, it 
is shown that the economist can predict and explain events with a modi- 
fied use of his conventional analytical tools. 

* The Journal of Political Economy, Vol. LVIII (1950), pp. 211-21. Reprinted by 
courtesy of the University of Chicago Press and the author. 

The author is indebted to Dr. Stephen Enke for criticism and stimulation leading 
to improvements in both content and exposition. 

t University of California at Los Angeles. 



This phenomenon — environmental adoption — is then fused with a 
type of individual motivated behavior based on the pervasiveness of un- 
certainty and incomplete information. Adaptive, imitative, and trial- 
and-error behavior in the pursuit of "positive profits" is utilized rather 
than its sharp contrast, the pursuit of "maximized profits." A final sec- 
tion discusses some implications and conjectures. 


Current economic analysis of economic behavior relies heavily on 
decisions made by rational units customarily assumed to be seeking per- 
fectly optimal situations. 1 Two criteria are well known — profit maximiza- 
tion and utility maximization. 2 According to these criteria, appropriate 
types of action are indicated by marginal or neighborhood inequalities 
which, if satisfied, yield an optimum. But the standard qualification usu- 
ally added is that nobody is able really to optimize his situation according 
to these diagrams and concepts because of uncertainty about the position 
and, sometimes, even the slopes of the demand and supply functions. 
Nevertheless, the economist interprets and predicts the decisions of in- 
dividuals in terms of these diagrams, since it is alleged that individuals 
use these concepts implicitly, if not explicitly. 

Attacks on this methodology are widespread, but only one attack has 
been really damaging, that of G. Tintner. 3 He denies that profit maximi- 
zation even makes any sense where there is uncertainty. Uncertainty 
arises from at least two sources: imperfect foresight and human inability 
to solve complex problems containing a host of variables even when an 
optimum is definable. Tintner's proof is simple. Under uncertainty, by 
definition, each action that may be chosen is identified with a distribution 
of potential outcomes, not with a unique outcome. Implicit in uncer- 
tainty is the consequence that these distributions of potential outcomes 
are overlapping. 4 It is worth emphasis that each possible action has a 
distribution of potential outcomes, only one of which will materialize if 
the action is taken, and that one outcome cannot be foreseen. Essentially, 
the task is converted into making a decision (selecting an action) whose 
potential outcome distribution is preferable, that is, choosing the action 

1 See, e.g., J. Robinson, Economics of Imperfect Competition (London: Mac- 
millan), p. 6, for a strong statement of the necessity of such optimal behavior. 
Standard textbooks expound essentially the same idea. See also P. Samuelson, 
Foundations of Economic Analysis (Cambridge: Harvard University Press, 1946). 

2 In the following we shall discuss only profit maximization, although everything 
said is applicable equally to utility maximization by consumers. 

3 "The Theory of Choice under Subjective Risk and Uncertainty," Econometrica, 
Vol. IX (1941), pp. 298-304; "The Pure Theory of Production under Technological 
Risk and Uncertainty," ibid., pp. 305-11; and "A Contribution to the Nonstatic 
Theory of Production," Studies in Mathematical Economics and Econometrics 
(Chicago: University of Chicago Press, 1942), pp. 92-109. 

4 Thus uncertainty is defined here to be the phenomenon that produces over- 
lapping distributions of potential outcomes. 


with the optimum distribution, since there is no such thing as a maxi- 
mizing distribution. 

For example, let each of two possible choices be characterized by its 
subjective distribution of potential outcomes. Suppose one has the higher 
"mean" but a larger spread, so that it might result in larger profits or 
losses, and the other has a smaller "mean" and a smaller spread. Which 
one is the maximum? This is a nonsensical question; but to ask for the 
optimum distribution is not nonsense. In the presence of uncertainty — 
a necessary condition for the existence of profits — there is no meaning- 
ful criterion for selecting the decision that will "maximize profits." The 
maximum-profit criterion is not meaningful as a basis for selecting the 
action which will, in fact, result in an outcome with higher profits than 
any other action would have, unless one assumes nonoverlapping poten- 
tial outcome distributions. It must be noticed that the meaningfulness 
of "maximum profits — a realized outcome which is the largest that could 
have been realized from the available actions" — is perfectly consistent 
with the meaninglessness of "profit maximization" — a criterion for select- 
ing among alternatives lines of action, the potential outcomes of which 
are describable only as distributions and not as unique amounts. 

This crucial difficulty would be avoided by using a preference func- 
tion as a criterion for selecting most preferred distributions of potential 
outcomes, but the search for a criterion of rationality and choice in 
terms of preference functions still continues. For example, the use of the 
mean, or expectation, completely begs the question of uncertainty by 
disregarding the variance of the distribution, while a "certainty equiva- 
lent" assumes the answer. The only way to make "profit maximization" 
a specifically meaningful action is to postulate a model containing cer- 
tainty. Then the question of the predictive and explanatory reliability 
of the model must be faced. 5 


There is an alternative method which treats the decisions and criteria 
dictated by the economic system as more important than those made by 
the individuals in it. By backing away from the trees — the optimization 
calculus by individual units — we can better discern the forest of imper- 
sonal market forces. 6 This approach directs attention to the interrelation- 
ships of the environment and the prevailing types of economic behavior 
which appear through a process of economic natural selection. Yet it 

5 Analytical models in all sciences postulate models abstracting from some realities 
in the belief that derived predictions will still be relevant. Simplifications are nec- 
essary, but continued attempts should be made to introduce more realistic assumptions 
into a workable model with an increase in generality and detail (see M. Friedman 
and L. Savage, "The Utility Analysis of Choices Involving Risks," ]ournal of 
Political Economy, Vol. LVI, No. 4 [1948], p. 279) . 

6 In effect, we shall be reverting to a Marshallian type of analysis combined with 
the essentials of Darwinian evolutionary natural selection. 


does not imply that individual foresight and action do not affect the 
nature of the existing state of affairs. 

In an economic system the realization of profits is the criterion accord- 
ing to which successful and surviving firms are selected. This decision 
criterion is applied primarily by an impersonal market system in the 
United States and may be completely independent of the decision proc- 
esses of individual units, of the variety of inconsistent motives and abili- 
ties, and even of the individual's awareness of the criterion. The reason 
is simple. Realized positive profits, not maximum profits, are the mark of 
success and viability. It does not matter through what process of reason- 
ing or motivation such success was achieved. The fact of its accomplish- 
ment is sufficient. This is the criterion by which the economic system 
selects survivors: those who realize positive profits are the survivors; 
those who suffer losses disappear. 

The pertinent requirement — positive profits through relative efficiency 
— is weaker than "maximized profits," with which, unfortunately, it has 
been confused. Positive profits accrue to those who are better than their 
actual competitors, even if the participants are ignorant, intelligent, skil- 
ful, etc. The crucial element is one's aggregate position relative to actual 
competitors, not some hypothetically perfect competitors. As in a race, 
the award goes to the relatively fastest, even if all the competitors loaf. 
Even in a world of stupid men there would still be profits. Also, the 
greater the uncertainties of the world, the greater is the possibility that 
profits would go to venturesome and lucky rather than to logical, careful, 
fact-gathering individuals. 

The preceding interpretation suggests two ideas. First, success (sur- 
vival) accompanies relative superiority; and, second, it does not require 
proper motivation but may rather be the result of fortuitous circum- 
stances. Among all competitors, those whose particular conditions hap- 
pen to be the most appropriate of those offered to the economic system 
for testing and adoption will be "selected" as survivors. Just how such an 
approach can be used and how individuals happen to offer these appro- 
priate forms for testing are problems to which we now turn. 7 


Sheer chance is a substantial element in determining the situation se- 
lected and also in determining its appropriateness or viability. A second 
element is the ability to adapt one's self by various methods to an ap- 
propriate situation. In order to indicate clearly the respective roles of 

7 Also suggested is another way to divide the general problem discussed here. 
The process and rationale by which a unit chooses its actions so as to optimize its 
situation is one part of the problem. The other is the relationship between changes 
in the environment and the consequent observable results, i.e., the decision process 
of the economic society. The classification used in the text is closely related to this 
but differs in emphasizing the degree of knowledge and foresight. 


luck and conscious adapting, the adaptive calculus will, for the moment, 
be completely removed. All individual rationality, motivation, and fore- 
sight will be temporarily abandoned in order to concentrate upon the 
ability of the environment to adopt "appropriate" survivors even in the 
absence of any adaptive behavior. This is an apparently unrealistic, but 
nevertheless very useful, expository approach in establishing the attenu- 
ation between the ex post survival criterion and the role of the individu- 
al's adaptive decision criterion. It also aids in assessing the role of luck 
and chance in the operation of our economic system. 

Consider, first, the simplest type of biological evolution. Plants "grow" 
to the sunny side of buildings not because they "want to" in awareness 
of the fact that optimum or better conditions prevail there but rather 
because the leaves that happen to have more sunlight grow faster and 
their feeding systems become stronger. Similarly, animals with configu- 
rations and habits more appropriate for survival under prevailing condi- 
tions have an enhanced viability and will with higher probability be 
typical survivors. Less appropriately acting organisms of the same general 
class having lower probabilities of survival will find survival difficult. 
More common types, the survivors, may appear to be those having 
adapted themselves to the environment, whereas the truth may well be 
that the environment has adopted them. There may have been no moti- 
vated individual adapting but, instead, only environmental adopting. 

A useful, but unreal, example in which individuals act without any 
foresight indicates the type of analysis available to the economist and 
also the ability of the system to "direct" resources despite individual 
ignorance. Assume that thousands of travelers set out from Chicago, se- 
lecting their roads completely at random and without foresight. Only 
our "economist" knows that on but one road are there any gasoline sta- 
tions. He can state categorically that travelers will continue to travel 
only on that road; those on other roads will soon run out of gas. Even 
though each one selected his route at random, we might have called 
those travelers who were so fortunate as to have picked the right road 
wise, efficient, foresighted, etc. Of course, we would consider them the 
lucky ones. If gasoline supplies were now moved to a new road, some 
formerly luckless travelers again would be able to move; and a new pat- 
tern of travel would be observed, although none of the travelers had 
changed his particular path. The really possible paths have changed with 
the changing environment. All that is needed is a set of varied, risk-tak- 
ing (adoptable) travelers. The correct direction of travel will be estab- 
lished. As circumstances (economic environment) change, the analyst 
(economist) can select the types of participants (firms) that will now 
become successful; he may also be able to diagnose the conditions most 
conducive to a greater probability of survival. 8 

8 The undiscerning person who sees survivors corresponding to changes in 
environment claims to have evidence for the "Lysenko" doctrine. In truth, all he 



These two examples do not constitute an attempt to base all analysis 
on adoptive models dominated by chance. But they do indicate that col- 
lective and individual random behavior does not per se imply a nihilistic 
theory incapable of yielding reliable predictions and explanations; nor 
does it imply a world lacking in order and apparent direction. It might, 
however, be argued that the facts of life deny even a substantial role 
to the element of chance and the associated adoption principle in the 
economic system. For example, the long lives and disparate sizes of busi- 
ness firms and hereditary fortunes may seem to be reliable evidence of 
consistent foresisfhted motivation and nonrandom behavior. In order to 
demonstrate that consistent success cannot be treated as prima facie evi- 
dence against pure luck, the following chance model of Borel, the fa- 
mous French mathematician, is presented. 

Suppose two million Parisians were paired off and set to tossing coins 
in a game of matching. Each pair plays until the winner on the first toss 
is again brought to equality with the other player. Assuming one toss per 
second for each eight-hour day, at the end of ten years there would still 
be, on the average, about a hundred-odd pairs; and if the players assign 
the game to their heirs, a dozen or so will still be playing at the end of a 
thousand years! The implications are obvious. Suppose that some busi- 
ness had been operating for one hundred years. Should one rule out luck 
and chance as the essence of the factors producing the long-term survival 
of the enterprise? No inference whatever can be drawn until the number 
of original participants is known; and even then one must know the size, 
risk, and frequency of each commitment. One can see from the Borel il- 
lustration the danger in concluding that there are too many firms with 
long lives in the real world to admit an important role to chance. On the 
contrary, one might insist that there are actually too few! 

The chance postulate was directed to two problems. On the one hand, 
there is the actual way in which a substantial fraction of economic be- 
havior and activity is effected. On the other, there is the method of 
analysis which economists may use in their predictions and diagnoses. 
Before modifying the extreme chance model by adding adaptive be- 
havior, some connotations and implications of the incorporation of 
chance elements will be elaborated in order to reveal the richness which 
is really inherent in chance. First, even if each and every individual 
acted in a haphazard and nonmotivated manner, it is possible that the 
variety of actions would be so great that the resulting collective set 
would contain actions that are best, in the sense of perfect foresight. 

may have is evidence for the doctrine that the environment, by competitive condi- 
tions, selects the most viable of the various phenotypic characteristics for perpetu- 
ation. Economists should beware of economic "Lysenkoism." 


For example, at a horse race with enough bettors wagering strictly at 
random, someone will win on all eight races. Thus individual random 
behavior does not eliminate the likelihood of observing "appropriate" 
decisions. 9 

Second, and conversely, individual behavior according to some fore- 
sight and motivation does not necessarily imply a collective pattern of 
behavior that is different from the collective variety of actions associated 
with a random selection of actions. Where there is uncertainty, people's 
judgments and opinions, even when based on the best available evidence, 
will differ; no one of them may be making his choice by tossing coins; 
yet the aggregate set of actions of the entire group of participants may 
be indistinguishable from a set of individual actions, each selected at 
random. 10 

Third, and fortunately, a chance-dominated model does not mean that 
an economist cannot predict or explain or diagnose. With a knowledge 
of the economy's realized requisites for survival and by a comparison of 
alternative conditions, he can state what types of firms or behavior rela- 
tive to other possible types will be more viable, even though the firms 
themselves may not know the conditions or even try to achieve them by 
readjusting to the changed situation if they do know the conditions. It is 
sufficient if all firms are slightly different so that in the new environ- 
mental situation those who have their fixed internal conditions closer to 
the new, but unknown, optimum position now have a greater probability 
of survival and growth. They will grow relative to other firms and be- 
come the prevailing type, since survival conditions may push the ob- 
served characteristics of the set of survivors toward the unknowable 
optimum by either (1) repeated trials or (2) survival of more of those 
who happened to be near the optimum — determined ex post. If these 
new conditions last "very long," the dominant firms will be different ones 
from those which prevailed or would have prevailed under other condi- 
tions. Even if environmental conditions cannot be forecast, the economist 
can compare for given alternative potential situations the types of be- 
havior that would have higher probability of viability or adoption. If 
explanation of past results rather than prediction is the task, the econo- 
mist can diagnose the particular attributes which were critical in facilitat- 
ing survival, even though individual participants were not aware of 
them. 11 

9 The Borel gamblers analogue is pertinent to a host of everyday situations. 

10 Of course, the economic units may be going through a period of soul searching, 
management training, and research activity. We cannot yet identify mental and 
physical activity with a process that results in sufficient information and foresight 
to yield uniquely determinate choices. To do so would be to beg the whole 

11 It is not even necessary to suppose that each firm acts as if it possessed the 
conventional diagrams and knew the analytical principles employed by economists 
in deriving optimum and equilibrium conditions. The atoms and electrons do not 


Fourth, the bases of prediction have been indicated in the preceding 
paragraph, but its character should be made explicit. The prediction will 
not assert that every — or, indeed, any — firm necessarily changes its char- 
acteristics. It asserts, instead, that the characteristics of the new set of 
firms, or possibly a set of new firms, will change. This may be character- 
ized by the "respresentative firm," a purely statistical concept — a vector 
of "averages," one dimension for each of the several qualities of the pop- 
ulation of firms. A "representative firm" is not typical of any one pro- 
ducer but, instead, is a set of statistics summarizing the various "modal" 
characteristics of the population. Surely, this was an intended use of 
Marshall's "representative firm." 

Fifth, a final implication drawn from consideration of this extreme ap- 
proach is that empirical investigations via questionnaire methods, so far 
used, are incapable of evaluating the validity of marginal productivity 
analysis. This is true because productivity and demand analyses are es- 
sential in evaluating reiative viability, even though uncertainty eliminates 
"profit maximization" and even if price and technological changes were 
to have no consciously redirecting effect on the firms. To illustrate, sup- 
pose that, in attempting to predict the effects of higher real wage rates, 
it is discovered that every businessman says he does not adjust his labor 
force. Nevertheless, firms with a lower labor-capital ratio will have rela- 
tively lower cost positions and, to that extent, a higher probability of 
survival. The force of competitive survival, by eliminating higher-cost 
firms, reveals a population of remaining firms with a new average labor- 
capital ratio. The essential point is that individual motivation and fore- 
sight, while sufficient, are not necessary. Of course, it is not argued here 
that therefore it is absent. All that is needed by economists is their own 
awareness of the survival conditions and criteria of the economic system 
and a group of participants who submit various combinations and organi- 
zations for the system's selection and adoption. Both these conditions 
are satisfied. 12 

As a consequence, only the method of use, rather than the usefulness, 
of economic tools and concepts is affected by the approach suggested 
here; in fact, they are made more powerful if they are not pretentiously 
assumed to be necessarily associated with, and dependent upon, individ- 
ual foresight and adjustment. They are tools for, at least, the diagnosis of 
the operation of an economic system, even if not also for the internal 
business behavior of each firm. 

know the laws of nature; the physicist does not impart to each atom a wilful scheme 
of action based on laws of conservation of energy, etc. The fact that an economist 
deals with human beings who have sense and ambitions does not automatically 
warrant imparting to these humans the great degree of foresight and motivations 
which the economist may require for his customary analysis as an outside observer 
or "oracle." The similarity between this argument and Gibbsian statistical mechanics, 
as well as biological evolution, is not mere coincidence. 

12 This approach reveals how the "facts" of Lester's dispute with Machlup can 
be handled with standard economic tools. 



Let it again be noted that the preceding extreme model was designed 
to present in purest form only one element of the suggested approach. 
It is not argued that there is no purposive, foresighted behavior present 
in reality. In adding this realistic element — adaptation by individuals with 
some foresight and purposive motivation — we are expanding the preced- 
ing extreme model. We are not abandoning any part of it or futilely try- 
ing to merge it with the opposite extreme of perfect foresight and "profit 

Varying and conflicting objectives motivate economic activity, yet we 
shall here direct attention to only one particular objective — the sufficient 
condition of realized positive profits. There are no implications of "profit 
maximization," and this difference is important. Although the latter is a 
far more extreme objective when definable, only the former is the sine 
qua JJon of survival and success. To argue that, with perfect competi- 
tion, the two would come to the same thing is to conceal an important 
difference by means of a very implausible assumption. The pursuit of 
profits, and not some hypothetical undefinable perfect situation, is the 
relevant objective whose fulfilment is rewarded with survival. Unfor- 
tunately, even this proximate objective is too high. Neither perfect 
knowledge of the past nor complete awareness of the current state of the 
arts gives sufficient foresight to indicate profitable action. Even for this 
more restricted objective, the pervasive effects of uncertainty prevent 
the ascertainment of actions which are supposed to be optimal in achiev- 
ing profits. Now the consequence of this is that modes of behavior re- 
place optimum equilibrium conditions as guiding rules of action. There- 
fore, in the following sections two forms of conscious adaptive behavior 
are emphasized. 

First, wherever successful enterprises are observed, the elements com- 
mon to these observable successes will be associated with success and 
copied by others in their pursuit of profits or success. "Nothing succeeds 
like success." Thus the urge for "rough-and-ready" imitative rules of 
behavior is accounted for. What would otherwise appear to be merely 
customary "orthodox," nonrational rules of behavior turns out to be codi- 
fied imitations of observed success, e.g., "conventional" markup, price 
"followship," "orthodox" accounting and operating ratios, "proper" ad- 
vertising policy, etc. A conventionally employed type of behavior pat- 
tern is consistent with the postulates of the analysis employed, even 
though the reasons and justifications for the particular conventions are 

13 These constructed rules of behavior should be distinguished from "rules" 
which, in effect, do no more than define the objective being sought. Confusion 
between objectives which motivate one and rules of behavior are commonplace. 
For example, "full-cost pricing" is a "rule" that one cannot really follow. He can 


Many factors cause this motive to imitate patterns of action observable 
in past successes. Among these are: (1) the absence of an identifiable 
criterion for decision making, (2) the variability of the environment, 
(3) the multiplicity of factors that call for attention and choice, (4) the 
uncertainty attaching to all these factors and outcomes, (5) the aware- 
ness that superiority relative to one's competitors is crucial, and (6) the 
nonavailability of a trial-and-error process converging to an optimum 

In addition, imitation affords relief from the necessity of really mak- 
ing decisions and conscious innovations, which, if wrong, become "in- 
excusable." Unfortunately, failure or success often reflects the willing- 
ness to depart from rules when conditions have changed; what counts, 
then, is not only imitative behavior but the willingness to abandon it at 
the "right" time and circumstances. Those who are different and success- 
ful "become" innovators, while those who fail "become" reckless vio- 
lators of tried-and-true rules. Although one may deny the absolute ap- 
propriateness of such rules, one cannot doubt the existence of a strong 
urge to create conventions and rules (based on observed success) and a 
willingness to use them for action as well as for rationalizations of inac- 
tion. If another untried host of actions might have been even more suc- 
cessful, so much the worse for the participants who failed, and even for 
those who missed "perfect success." 

Even innovation is accounted for by imitation. While there certainly 
are those who consciously innovate, there are those who, in their imper- 
fect attempts to imitate others, unconsciously innovate by unwittingly 
acquiring some unexpected or unsought unique attributes which under 
the prevailing circumstances prove partly responsible for the success. 
Others, in turn, will attempt to copy the uniqueness, and the imitation- 
innovation process continues. Innovation is assured, and the notable as- 
pects of it here are the possibility of unconscious pioneering and leader- 

The second type of conscious adaptive behavior, in addition to imita- 
tion, is "trial and error." This has been used with "profit maximization," 
wherein, by trial and ensuing success or failure, more appropriate actions 
are selected in a process presumed to converge to a limit of "profit maxi- 
mization" equilibrium. Unfortunately, at least two conditions are neces- 
sary for convergence via a trial-and-error process, even if one admits an 
equilibrium situation as an admissible limit. First, a trial must be classi- 
fiable as a success or failure. The position achieved must be comparable 
with results of other potential actions. In a static environment, if one im- 
proves his position relative to his former position, then the action taken 

try to, but whether he succeeds or fails in his objective of survival is not controllable 
by following the "rule of full-cost pricing." If he fails in his objective, he must, 
of necessity, fail to have followed the "rule." The situation is parallel to trying to 
control the speed of a car by simply setting by hand the indicator on the speedometer. 


is better than the former one, and presumably one could continue by 
small increments to advance to a local optimum. An analogy is pertinent. 
A nearsighted grasshopper on a mound of rocks can crawl to the top of 
a particular rock. But there is no assurance that he can also get to the 
top of the mound, for he might have to descend for a while or hop to 
new rocks. The second condition, then, for the convergence via trial 
and error is the continual rising toward some optimum optimorum with- 
out intervening descents. Whether decisions and actions in economic life 
satisfy these two conditions cannot be proved or disproved here, but the 
available evidence seems overwhelmingly unfavorable. 

The above convergence conditions do not apply to a changing en- 
vironment, for there can be no observable comparison of the result of 
an action with any other. Comparability of resulting situations is de- 
stroyed by the changing environment. As a consequence, the measure of 
goodness of actions in anything except a tolerable-intolerable sense is 
lost, and the possibility of an individual's converging to the optimum 
activity via a trial-and-error process disappears. Trial and error becomes 
survival or death. It cannot serve as a basis of the individual's method of 
convergence to a "maximum" or optimum position. Success is discovered 
by the economic system through a blanketing shotgun process, not by 
the individual through a converging search. 

In general, uncertainty provides an excellent reason for imitation of 
observed success. Likewise, it accounts for observed uniformity among 
the survivors, derived from an evolutionary, adopting, competitive sys- 
tem employing a criterion of survival, which can operate independently 
of individual motivations. Adapting behavior via imitation and venture- 
some innovation enlarges the model. Imperfect imitators provide 
opportunity for innovation, and the survival criterion of the economy 
determines the successful, possibly because imperfect, imitators. Innova- 
tion is provided also by conscious wilful action, whatever the ultimate 
motivation may be, since drastic action is motivated by the hope of great 
success as well as by the desire to avoid impending failure. 

All the preceding arguments leave the individual economic participant 
with imitative, venturesome, innovative, trial-and-error adaptive behav- 
ior. Most conventional economic tools and concepts are still useful, al- 
though in a vastly different analytical framework — one which is closely 
akin to the theory of biological evolution. The economic counterparts of 
genetic heredity, mutations, and natural selection are imitation, innova- 
tion, and positive profits. 


I shall conclude with a brief reference to some implications and con- 

Observable patterns of behavior and organization are predictable in 
terms of their relative probabilities of success or viability if they are 


tried. The observed prevalence of a type of behavior depends upon both 
this probability of viability and the probability of the different types 
being submitted to the economic system for testing and selecting. One 
is the probability of appearance of a certain type of organization (muta- 
tion), and the other is the probability of its survival or viability, once it 
appears (natural selection). There is much evidence for believing that 
these two probabilities are interrelated. But is there reason to suppose 
that a high probability of viability implies a high probability of an ac- 
tion's being taken, as would be implied in a system of analysis involving 
some "inner directed urge toward perfection"? If these two probabili- 
ties are not highly correlated, what predictions of types of action can the 
economist make? An answer has been suggested in this paper. 

While it is true that the economist can define a profit maximization 
behavior by assuming specific cost and revenue conditions, is there any 
assurance that the conditions and conclusions so derivable are not too 
perfect and absolute? If profit maximization (certainty) is not ascertain- 
able, the confidence about the predicted effects of changes, e.g., higher 
taxes or minimum wages, will be dependent upon how close the formerly 
existing arrangement was to the formerly "optimal" (certainty) situation. 
What really counts is the various actions actually tried, for it is from these 
that "success" is selected, not from some set of perfect actions. The 
economist may be pushing his luck too far in arguing that actions in 
response to changes in environment and changes in satisfaction with the 
existing state of affairs will converge as a result of adaptation or adop- 
tion toward the optimum action that should have been selected, if fore- 
sight had been perfect. 14 

In summary, I have asserted that the economist, using the present 
analytical tools developed in the analysis of the firm under certainty, can 
predict the more adoptable or viable types of economic interrelation- 
ships that will be induced by environmental change even if individuals 
themselves are unable to ascertain them. That is, although individual 
participants may not know their cost and revenue situations, the econo- 
mist can predict the consequences of higher wage rates, taxes, govern- 

14 An anomalous aspect of the assumption of perfect foresight is that it nearly 
results in tautological and empty statements. One cannot know everything, and this 
is recognized by the addendum that one acts within a "given state and distribution 
of the arts." But this is perilously close, if not equivalent, to saying either that 
action is taken only where the outcome is accurately foreseen or that information 
is always limited. The qualification is inserted because one might contend that it 
is the "constancy of the state and distribution of arts" that is necessary as a ceteris 
paribus. But even the latter is no solution. A large fraction of behavior in a world of 
incomplete information and uncertainty is necessarily directed at increasing the 
state of arts and venturing into an unknown sphere. While it is probably permissible 
to start with a prescribed "distribution of the knowledge of the arts," holding it 
constant is too restrictive, since a large class of important and frequent actions 
necessarily involves changes in the state and distribution of knowledge. The modifi- 
cation suggested here incorporates this search for more knowledge as an essential 


ment policy, etc. Like the biologist, the economist predicts the effects of 
environmental changes on the surviving class of living organisms; the 
economist need not assume that each participant is aware of, or acts ac- 
cording to, his cost and demand situation. These are concepts for the 
economist's use and not necessarily for the individual participant's, who 
may have other analytic or customary devices which, while of interest 
to the economist, serve as data and not as analytic methods. 

An alternative to the rationale of individual profit maximization has 
been presented without exorcising uncertainty. Lest isolated arguments 
be misinterpreted, let it be clearly stated that this paper does not argue 
that purposive objective-seeking behavior is absent from reality, nor, on 
the other hand, does it indorse the familiar thesis that action of economic 
units cannot be expressed within the marginal analysis. Rather, the con- 
tention is that the precise role and nature of purposive behavior in the 
presence of uncertainty and incomplete information have not been 
clearly understood or analyzed. 

It is straightforward, if not heuristic, to start with complete uncer- 
tainty and nonmotivation and then to add elements of foresight and moti- 
vation in the process of building an analytical model. The opposite ap- 
proach, which starts with certainty and unique motivation, must abandon 
its basic principles as soon as uncertainty and mixed motivations are rec- 
ognized. 15 The approach suggested here is intellectually more modest and 
realistic, without sacrificing generality. It does not regard uncertainty as 
an aberrational exogenous disturbance, as does the usual approach from 
the opposite extreme of accurate foresight. The existence of uncertainty 
and incomplete information is the foundation of the suggested type of 
analysis; the importance of the concept of a class of "chance" decisions 
rests upon it; it permits of various conflicting objectives; it motivates 
and rationalizes a type of adaptive imitative behavior; yet it does not 
destroy the basis of prediction, explanation, or diagnosis. It does not base 
its aggregate description on individual optimal action; yet it is capable of 
incorporating such activity where justified. The formalization of this ap- 
proach awaits the marriage of the theory of stochastic processes and eco- 
nomics — two fields of thought admirably suited for union. It is conjec- 
tured that the suggested modification is applicable to a wide class of 
events and is worth attempts at empirical verification. 16 

15 If one prefers, he may believe that the suggestions here contain reasons why 
the model based on certainty may predict outcomes, although individuals really 
cannot try to maximize profits. But the dangers of this have been indicated. 

16 Preliminary study in this direction has been very convincing, and, in addition, 
the suggested approach appears to contain important implications relative to general 
economic policy; but discussions of these are reserved for a later date. 


A Note on Pricing in Monopoly 
and Oligopoly* 


The conventional versions of a priori price analysi s ap parently sug- 
gest that a single-firm monopoly or a collusive oligopoly will choose a 
price-output combination such as to maximize the industry profit. Prod- 
uct differentiation and selling cost being neglected, the currently estab- 
lished firm or firms are supposed _to_ equate their marginal costs to the 
marginal revenue drawn fro m the industry demand curve for the com- 
modity which they produce in commorLjThis balance should presumably 
be struck over any long period between long-run marginal cost and the 
marginal revenue from the long-run industry demand, and in any short 
period between short-run marginal cost and the marginal revenue from 
the short-run industry demand. Price in either period should be set to 
maximize the difference between the aggregate revenue from the sale of 
the given commodity and the aggregate cost of its production by anv 
established group of firms. Empirical studies of pric e policy by mo- 
nopo lists or b^_o^gg^^^j^h^ Ml$MM^ effective collusion on p rice, 
however, frequently fail ._sustain_d^sejgr£dictions. In many sucrTTn- 
dustries. short-run o utputs at which short-run margin al costs plainly~ex- 
ceed short-run industry marginal revenue are apparently common. bu t~ 
more striking is the evidence in some of these industries of prices held 
persistently over many years within a ran ge~where the industry demand" 
curve is evidently inelastic, the ^^ correspondingmarginal"Teyenue thus be- 
ing negative and necessarily below long-run marginal cost. 1 This indi- 
cates a prolonged tendency (potentially for a theoretical "long run^)__to 
hold price well below the level which would maximize the difference 
b etwe en aggregate re venue from the sale of the industry's commodity 
and the aggregate cost of producing it, 2 and apparently contradicts the 
basic a priori predictions of a theory of collusive pricing. 

* The American Economic Review, Vol. XXXIX (1949), pp. 448-64. Reprinted 
by courtesy of the publisher and the author. 

The author is indebted to Professors H. S. Ellis, W. Fellner, and R. A. Gordon 
for helpful criticism of the paper and for a number of substantive suggestions. 
I University of California. 

1 Two fairly convincing examples of this are the cigarette and steel industries. 

2 Of producing it, explicitly, with the "given" number of firms, but also with 
any other conceivable number of firms if industry marginal revenue is negative. 



This apparent impasse has been variously resolved by students of in- 
dustry with suggestions: (a) that sellers do not try to maximize monetary 
profit; (b) that they err in their attempt to maximize profits; (c) that in 
the face of great uncertainty concerning demand they simply add some 
markup to normal average cost and hope for the best; (d) that they fear 
government interference and public ill will if they exploit their monop- 
oly positions fully; (e) that the apparently collusive oligopoly is not 
fully or successfully collusive, so that rivalry keeps price down; (f) that 
sellers set low prices for very considerable time periods in order to raise 
the level of future industry demand; and (g) that established sellers per- 
sistently or "in the long run" forego prices high enough to maximize the 
industry profit for fear of thereby attracting new entry to the industry 
and thus reducing the demands for their outputs and their own profits. 

Although each of these explanations may contain an element of truth 
as applied to particular cases, they proceed on somewhat different levels 
in their implied criticisms of conventional theory. The rejection of profit 
maximization as a goal suggests a corresponding rejection of conven- 
tional price theory; the thesis that either errors or uncertainties are dom- 
inant suggests that if such theory is basically valid in its assumptions it 
nevertheless has little genuine value for predicting actual price results. 
The suggestion that oligopolistic rivalry reduces prices below the mo- 
nopoly level leaves theory unscathed, simply implying that the model 
for monopoly pricing has been misapplied. On the other hand, the hy- 
potheses concerning fear of interference, threat of entry, and pricing to 
stimulate future demand do not deny that the observed results may be 
explicable in terms of a theory of monopoly or collusive oligopoly price 
which assumes profit maximization, effective collusion, and approxi- 
mately given data, but suggest that the industry or seller demand curves 
employed in that theory must be redrawn to reflect explicitly the effects 
of the phenomena in question. 

Until the results of such a redrafting are explored, we cannot properly 
assess the potential explanatory value of conventional price theory. In 
this paper, we examine two possible modifications of the theory of mo- 
nopoly price, to take account first of the relation of present price to 
future profit and second of the impact of the threat of entry; 3 in the 
latter case we suggest certain possibly novel conclusions concerning the 
unit for which profit may be maximized and the relation of marginal 
cost to marginal revenue. 


An elaboration of conventional theory to recognize the relation of 
current price to future demand has been suggested by M. W. Reder; it 
may deserve re-emphasis in connection with the present issue. 4 

3 Although the threat of government interference will not be treated explicitly, 
it could be handled in about the same manner as entry. 

4 "Intertemporal Relations of Demand and Supply within the Firm," Canadian 


The monopolistic firm (or group of collusive oligopolists 5 ) may be 
provisionally viewed as dealing with an entire industry demand curve in 
a succession of time intervals, in each of which it can freely select a 
price-output combination for the industry. It will thus logically take 
account of the effect of any current price-output decision on the position 
of the industry demand curve in future periods. A lower price now may 
mean a larger (or smaller) demand later, and any such anticipated re- 
lationship should affect any current pricing policy. If it does, a single 
long-run industry demand curve cannot be viewed as an independent 
determinant even of the long-run tendency of price. Such a relation is 
then not given independently of the prices which the seller (s) charge 
at various times during the future, but will assume various levels ac- 
cording to the behavior of a series of short-run prices. Instead the sell- 
er(s) necessarily refer to a series of short-period demand curves for each 
of a succession of future intervals; these fully replace any long-run de- 
mand curve for the purposes of making all output adjustments. And we 
must now speak not of a single long-run tendency for price, but rather 
of a price-pattern through time. 

A simple model can be constructed in which a single-firm monopolist 
is conceived of as pricing solely in a current Period I and a future Pe- 
riod II. There is an industry demand curve for each such period and a 
corresponding pair of marginal revenue curves. Monopoly price for the 
first period in isolation would be set to equate the marginal cost of that 
period to the marginal revenue of that period. But demand in Period II 
may be supposed to depend upon Period I price; for example, the Pe- 
riod II demand curve may shift outward as Period I price falls. Viewing 
this relation in prospect from the beginning of the first period, the mo- 
nopolist may be supposed to adjust Period I price so as to allow maxi- 
mization of the sum of the profits of the two periods. This procedure 
may result in setting Period I price below the level for which the mar- 
ginal cost and marginal revenue of that period are equated, so long as 
the resulting decrement to Period I profits is more than offset by a re- 
sulting increment in Period II profits (as appropriately discounted for 
interest and risk). 

Use of the demand-supply technique in such a sequence analysis per- 
mits precise formal treatment of the effects of anticipated relations be- 
tween current price and future demand. To employ the analysis for 
purposes of prediction, we should attempt to determine the sign and 
value of the cross-elasticity between Period I price and Period II quan- 
tity for the monopolist — 

dpu=» #<*=» 
Journal of Economics and Political Science (February, 1941), pp. 25-38, and es- 
pecially pp. 32-35. 

5 For a simplified argument, we will suppose the collusive oligopoly in each 
case to be a "pure" oligopoly, in which the several firms sell identical products at a 
necessarily identical price (the possibility of discrimination being neglected). 


One can identify actual cases where this elasticity might be alternatively 
positive or negative in sign, or zero. Where it is negative and signifi- 
cantly large, a current monopoly price below the current profit-maxi- 
mizing level can be formally explained; where it is positive and large, 
current prices above this level could be predicted. This model of course 
does not have the monopoly doing other than to maximize the "long- 
run" difference between aggregate industry revenue and aggregate pro- 
duction cost, so long as these are measured as capital values of future 
revenue and cost streams to a time horizon. But it does indicate the 
rational possibility of deliberate departures from profit maximization for 
"short" periods longer than those required to permit every adaptation in 
scale of firm and plant. 


Let us now turn to the effects of anticipated entry. Even a single-firm 
monopoly is not necessarily impregnable to entry if the industry is a 
very profitable one, and in oligopolistic industries the threat of entry is 
likely to be stronger. The monopolist or the group of collusive oligopo- 
lists might therefore be viewed as setting each of an indefinite succession 
of current prices or profits with an eye to their effect in attracting entry 
into the industry and thus in reducing the demand for the output of the 
now-established firm(s). 6 One possibility is that the initially established 
seller or collusive sellers will be faced with the choice of (a) setting each 
of a succession of short-run prices (and hence long-run average price) 
so as to maximize the industry profit, but with the result that added 
firms enter the industry and reduce the share of industry profit gained 
by the initially established firms, and (b) setting each short-run price 
(and hence long-run average prices) at a lower level, thus discouraging 
further entry, and keeping the smaller (and non-maximized) industry 
profit all for themselves. Should the second course then offer larger long- 
run profits to the initially established firms, they would presumably fol- 
low it, and price could for the indefinitely long run lie below the level 
required to maximize the difference between the aggregate revenue from 
the sale of the given commodity and the aggregate cost of producing it. 
Long-run maximization of industry profit and of the profit of a group of 
currently established firms may not coincide. This is presumably a thesis 
implied by those who point to threat of entry as a factor holding price 
below the level which would maximize the long-run profits of the in- 

This hypothesis can easily be developed on a formal level if we accept 
two premises upon which it must implicitly be based. These are ( 1 ) that 
the established monopolist or group of collusive oligopolists are aware 
of any real threat of entry to their industry, and will adjust to it in such 

6 The threat of entry, and its relation to price, will of course depend upon the 
height and effectiveness of such institutional barriers as patent holdings, control of 
raw materials, etc. 


a way as to enhance their own (as distinguished from industry) profits, 
and (2) that potential entrants to such a monopoly or oligopoly are pri- 
marily influenced in deciding whether or not to enter by the prices 
charged and profits currently earned by the established firms. The first 
premise seems only reasonable, 7 but the merit of the second could be 
contested. A potential entrant to a purely competitive industry is pre- 
sumably guided entirely by the expected long-run tendency of industry 
price as related to his contemplated costs. This is because his increment 
to the industry would neither perceptibly influence the price nor en- 
gender any direct reaction from any established seller. The potential 
entrant to a monopoly or oligopoly, on the other hand, who will typically 
make a substantial lump increment to the industry capacity (since econ- 
omies of scale will ordinarily require a fairly large firm), may expect 
both to influence the pre-existing price and to elicit some reaction from 
the established seller (s). This holds whether or not he contemplates col- 
lusion with those sellers. In effect, there is a special sort of oligopolistic 
interdependence between the established seller (s) and the potential en- 
trants in such instances, and it is not entirely plausible that the potential 
entrant should entirely neglect this interdependence and view the going 
industry price as the principal indicator of whether or not entry will be 

At the extreme, it could even be argued that a potential entrant to an 
oligopoly should pay little regard to price or profit received by estab- 
lished firms, especially if he thought price was being held down in order 
to "bluff" him away from the industry. He should look at the industry 
demand, the current competitive or collusive conditions in the industry, 
the prospects for rivalry or collusion after his entry, the share of the 
market he expects to capture, and his projected costs of production. 
Paramount in his considerations, provided the industry demand under 
some conceivable arrangement could provide profits to an entrant, should 
be his appraisal of the sort of rivalry and the type of price policies he 
will encounter from the previously established seller (s) after he enters. 
In judging these determinants of his decision, current price or profit in 
the industry need play no direct role, since the anticipated industry price 
after entry and the entrant's anticipated market share are the strategic 
considerations. And if he knows the industry demand with reasonable 
certainty and makes calculations concerning the conditions of rivalry 
after his entry, upon which he is willing to act, he might look entirely 
past any current price set by the established firm(s). He then would be 
immune to bluffing, and the established firm(s) could never discourage 
entry by lowering prices and earning moderate profits. 

The supposition that the potential entrant's judgment of industry de- 

7 To argue that sellers in concentrated industries deliberately disregard the 
consequences of threatened entry would picture them as unbelievably stupid. 


mand and of the rivalry he will meet is entirely unrelated to current 
price or profit in the industry, however, probably goes too far. Even if 
he does not believe the observed price will remain there for him to ex- 
ploit, he may nevertheless regard this price as an indicator both of the 
character of industry demand and of the probable character of rival 
policy after his entry. Industry demands are never certainly known, and 
they are probably known less fully by potential entrants than by estab- 
lished firms. The fact that the established firm(s) make only moderate 
profits may thus create in the mind of the potential entrant sufficient 
uncertainty concerning the elasticity of the industry demand curve at 
higher prices to deter him from entering. Moreover, he may view the 
price which the established firm(s) currently charge as a partial indicator 
of the rival price policy he will face after entry. Other considerations 
should influence his judgment of projected rivalry, but current pricing 
may be a critical factor in evaluating it. It is thus possible that the po- 
tential entrant is influenced by current prices and profits and that there 
may be a critical price below which he will not enter and above which 
he will enter. This hypothesis seems plausible enough so long as the po- 
tential entrant regards the current pricing policy of established sellers 
as being probably a statement of intentions rather than a bluff. It is 
probably more plausible than otherwise when applied to oligopolies 
where product differentiation is not very great, and where the entry 
problem is thus not unduly complicated by the necessity of gaining 
buyer acceptance of a new product. We will speak here primarily of its 
application to oligopolies with relatively slight product differentiation. 8 
On the basis of the preceding argument, we may provisionally accept 
the second premise of the thesis concerning threat of entry, that the 
potential entrant to a monopoly or oligopoly is primarily influenced by 
the price charged (and profit earned) by the established seller(s) — in- 
fluenced not because he expects this price to hold unchanged after entry, 
but because he regards it as "proving" the industry demand at a given 
level and as a critical indicator of the projected state of rivalry or price 
policy after entry. Accepting the first and second premises, we may now 
investigate their formal implications for the price policy of an estab- 
lished monopolist or group of collusive (pure) oligopolists faced with 
a threat of entry. What will happen in these cases if (a) the established 
seller (s) anticipate any threat of entry (highly probable), (b) potential 
entrants are influenced in their entry decisions by current price in the 
industry (a strong possibility), and (c) the established seller (s) know 
this and consider adjusting their prices to discourage entry. 

8 It may be added that with imperfect market information, "potential entrants" 
may be made aware of the possibilities of an industry mainly by its profit record, 
and will never present an active threat if they are not alerted by high profits. In 
this case, low current prices and profits may serve as a deterrent to entry because 
they do not attract attention and thereby create actively potential entrants. 



For a formal treatment, two special concepts may be conveniently 
employed. The first is the "limit price," or highest common price which 
the established seller (s) believe they can charge without inducing at 
least one increment to entry — presumably a significant lump increment. 
This limit price depends ultimately upon the cost functions which po- 
tential entrants expect to have, upon their estimates of the industry de- 
mand and of the share of the market which they can capture if they 
enter, and upon their view of the degree of competition or collusion 
which will obtain in the industry after their entry. The subjective esti- 
mate of this limit price by the established seller (s), however, rather than 
the view of potential entrants, is the real determinant of the price policies 
of the established seller (s). Since the limit price must be defined in terms 
of the guess of the established firm(s) concerning the anticipations of 
the potential rivals, it is especially subject to error as an ex ante mag- 
nitude, and it may be invalid if potential entrants read it as a bluff. But 
it is nevertheless potentially valid and determinate. 9 The second concept 
is the estimate by the established seller (s) of the conditions of demand 
for their outputs after entry occurs in response to their setting a price 
above the limit. This involves their estimate of the market share they will 
lose to an entrant, and also of the conditions of competition or collusion 
which will obtain after entry. If the established seller (s) formulate esti- 
mates on limit price, as defined, and on the position and character of 
the demand for their outputs after entry, these estimates can be recog- 
nized in an anticipated demand curve or sequence of demand curves for 
their outputs, and a formal solution developed. 

This solution can follow the lines developed for the case of interde- 
pendence of prices through time. We can first construct a given industry 
demand curve for the current Period I and for Periods II, III, etc. 10 The 
demand for the output of the established seller (s) in the current Period I 
is the industry demand of that period. In any later period, however, it 
can be expected to be the same as the industry demand only so long as 
price in the preceding period has remained at or below a limit price, A, 
so that entry has been forestalled. If any given period price is set above 
A, the demand for the output of the established seller(s) in all later 
periods is expected to become less than the industry demand by the 
amount of a market share going to the new entrant, and it may be other- 
wise altered or made uncertain if effective collusion with the entrant is 
not contemplated or considered attainable. If entry will occur in discrete 
lumps, moreover, the demand for the output of the established seller (s) 

9 Account must be taken, of course, of the consequences of erroneous as well 
as correct estimates of this price by the established seller (s). 

10 Each such industry demand curve is assumed to be independently given in 
the absence of interperiod price relationships. 


in later periods does not shift continuously in response to variations in 
Period I price. Instead it makes a discrete shift backward if A is sur- 
passed in Period I, and the solution is affected by this discontinuity. 
Given these conditions, the established seller (s) will devise a price policy 
for Period I and in all later periods so as to maximize the discounted 
present value of profits for all future periods. With an effective threat 
of entry, it is potentially consistent with such profit maximization by 
the established seller (s) that price will be held at the limit level con- 
tinually through time, even though this limit price may in every current 
period be lower than that for which the marginal cost of the established 
seller (or horizontally added marginal costs of the established sellers) 
equals the marginal revenue drawn from the industry demand curve for 
that period. Such a solution can be determinate and give stability with- 
out entry for the indefinitely long run (permitting all desired adjust- 
ments of scale by the existing firms) provided the limit price estimated 
by the established seller(s) is in fact low enough to exclude entry. If 
the established seller (s) set a "limit" price which turns out to be too high 
to exclude entry, of course, their error may result in an effectively ir- 
reversible change in the structure of the industry. 

Because the sequence analysis is unnecessarily awkward for dealing 
with the threat of entry, the preceding solution is only sketched. As- 
suming no interdependence of the industry demands of successive time 
periods to be involved in the case, the impact of a threat of entry can 
be analyzed more easily by referring to the anticipated long-run demand 
conditions for the output of the established monopolist or collusive oli- 
gopolists — or to the expected response of their long-run average sales to 
changes in the average level of price they maintain over long periods. 
Following this course, two alternative models may be developed, one 
assuming that the established seller (s) anticipate rivalry and lack of 
agreement with any new entrant, and the other assuming that they antici- 
pate collusion with any new entrant. 

The first model postulates (1) a determinate long-run demand curve 
for industry output, which is unaffected by price adjustments or by 
entry; (2) occupation of the industry initially by a single-firm monopo- 
list or group of effectively collusive pure oligopolists; (3) estimation by 
the established seller (s) of a limit price above which a "lump" of entry 
will be attracted; and (4) considerable uncertainty on the part of the es- 
tablished seller (s) concerning the conditions of demand for their outputs 
if entry is attracted. Given these conditions, the anticipated long-run 
demand for the output of the established seller (s) may be analyzed as 
follows. The long-run industry demand curve is supposed to be the line 
DABD', as in Figure 1, and the marginal revenue drawn to it is Dab?n. 
This demand is assumed to be unaffected by any adjustments sellers may 
make. Suppose now that the limit price above which the established 
seller (s) expect an increment of entry is Q a A: If they charge more than 



this (or produce less output than OQ a , thus causing the effective market 
price to exceed the limit), they expect to experience some indeterminate 
loss in sales volume to an entrant and some indeterminate change in 
price. The anticipated demand curve for the output of the established 
seller (s) above the price Q a A thus is not DA. They have the truncated 
demand curve AD' to exploit up to the price Q a A, and the corresponding 

Figure 1 

marginal revenue segment am. But if they raise or attempt to raise long- 
run price above Q a A, the anticipated long-run demand curve for their 
output becomes indeterminate somewhere in the range to the left of A. 
They thus have the choice of the truncated industry demand curve AD' 
for exclusive exploitation up to the price Q a A, and an indeterminate 
output becomes indeterminate somewhere in the range to the left of A. 
They thus have the choice of the truncated industry demand curve AD' 
for exclusive exploitation up to the price Q a A, and an indeterminate 
demand for their outputs if they once go above Q a A. It should be es- 
pecially noted that they are unable to sell less than the amount OQ a at 
the price Q a A and thus to exclude entry, since this would result in an 


effective market price higher than the limit, via resales, and thus pre- 
sumably "reveal the bluff" and attract entry. 11 

The alternatives open to the established seller (s) are now (1) to sell 
more than OQ a at a price below Q a A, thus excluding entry, (2) to sell 
OQ a at the price Q a A y also excluding entry, and (3) to raise price above 
Q a A, or reduce output below OQ a , thus attracting entry and taking 
chances on profits and prices in the ensuing indeterminate situation. They 
will presumably pursue the course that promises to be most profitable, 
taking account of the fact that profits under courses (1) and (2) are 
relatively determinate whereas profits under course (3) are indetermi- 
nate and hence highly uncertain. The established seller (s) will follow 
the first or second course in preference to the third wherever the rela- 
tively certain profits offered by those courses exceed the heavily risk- 
discounted gain attainable if entry is attracted via higher prices. The 
possible positions in which the established seller (s) in various industries 
may find equilibrium may be illustrated as follows. Suppose a single 
long-run marginal cost curve, for the initially established monopolist, or 
alternatively a uniquely determined aggregation of marginal cost curves 
for the initially established collusive oligopolists in any industry. This 
we label MC. Now first this marginal cost may in some industries lie 
at MCi (Figure 1), intersecting the relevant industry marginal revenue 
segment am. 12 Then the established seller (s) will almost certainly set 
price and output by this intersection, provided average costs are less 
than the resulting price. In this case, industry profit will be maximized 
at a price below the limit, entry will be forestalled, and the number of 
sellers in the industry will be in long-run equilibrium. Conventional mo- 
nopoly maximization is possible without further entry being attracted. 
This case subsumes all those where entry is blockaded or where the 
limit price is so high as to be economically irrelevant. 

Second, marginal cost may in other industries fall at MC%, lying above 
industry marginal revenue but below price at the limit output OQ a , with 
average costs less than Q a A. In this case, provided the profit offered 
seems preferable to the gamble of inducing entry, the established sell- 
er(s) will produce OQ a and sell at Q a A. (They will then not choose the 
intersection of MC2 and the marginal revenue D a , since this would give 
a price which would induce entry.) In this case entry is also forestalled 
and the number of sellers in the industry is in long-run equilibrium, but 

11 We exclude herewith the possibility of effective private rationing or price 
discrimination by the established seller (s), which might enable them to produce 
less than OQ a and still hold the effective market price at Q a A; this appears to be 
a special and unlikely case. 

12 In this and each of the succeeding cases we refer to distinct industry situations, 
each with a separate limit price, a separate initial marginal and average cost function, 
and a different relation of marginal and average cost to the limit price. We do not 
suggest different relations of cost to limit price in a single industry, but rather 
differences among industries in this respect. 


marginal cost exceeds industry marginal revenue and industry profits are 
not maximized. 

Third, marginal cost may fall at AfC 3 , lying above price at the limit 
output OQ a , but with the corresponding average cost lying below price 
at this output. The established sellers will still choose to produce OQ a 
and sell at Q a A, so long as the resulting profit is considered preferable 
to the gamble if entry is induced. Again the number of sellers will re- 
main constant, but industry profits will not be maximized and marginal 
cost will exceed price — not a probable but nevertheless a quite possible 
and rational result. 

The general argument developed for the last two cases may also be 
applied on the supposition that the limit price lies at some level Q h B on 
the industry demand curve, where this demand is less elastic than unity 
and the relevant marginal revenue segment, bm, is entirely in the nega- 
tive range. We may still have equilibrium with entry forestalled at the 
limit price Q b B (not rationally below it) with marginal cost above in- 
dustry marginal revenue and possibly above price, but with the differ- 
ence that industry marginal revenue is negative. 

These solutions involve the premises that potential entrants recognize 
a limit price below which they will not enter, and that the established 
seller (s) know this and do not overestimate the limit price. Should po- 
tential entrants fail to be influenced by price, a stable solution will not 
result if entry promises to be profitable; if they are so influenced but the 
established seller (s) set too high a price, there will be entry and a prob- 
ably irreversible change in industry structure will result. One qualifica- 
tion may be added to the preceding argument. In deciding whether or 
not to go above the limit price, the established sellers should count in 
favor of the former course any transitional extra profit they may earn 
after going above the limit and before entry becomes effective. 13 This 
consideration has not been formally treated in the preceding model. 

A fourth possibility is that average cost for the established seller (s) 
will lie above the limit price Q a A at the output OQ a , marginal cost lying 
above or below price. In this event price will presumably be set above 
the limit and entry attracted, provided there is some possibility of mak- 
ing profits at smaller outputs. 14 The number of sellers in the industry 
then could not be stable until further entry had occurred. 

Considering the various possibilities, there is a very good a priori 
chance on the assumptions drawn for the threat of entry to force a 
significant departure from what have been viewed as the conventional 

13 In a dynamic model, we might consider the possibility of a critical or maximum 
short period during which established sellers could temporarily go above limit price 
without attracting entry, returning price to the limit in time to discourage potential 

14 In this case, however, the potential entrants presumably being able to attain 
lower average costs than established firms, it is doubtful that any price stratagem 
would forestall entry. 


long-run monopoly-equilibrium price and output in single-firm monop- 
oly or collusive oligopoly industries. It has been conventionally supposed 
that the single-firm monopolist will set a price such as to maximize long- 
run industry profit, and that collusive oligopolists will do likewise. It 
has been further suggested that this may result, at least in collusive oli- 
gopoly, in the attraction of entry to the point where excess profits are 
small or absent and the industry contains an excessive number of firms. 15 
But under the assumptions of anticipated entry and a response of entry 
to price, we see that, consistent with profit maximization by firms, the 
price in such industries may be lower and the output larger than would 
maximize long-run industry profit. A vigorous threat of entry which at 
an appropriate time is anticipated and forestalled, moreover, may serve 
to keep firms producing at outputs which give a fairly close approxima- 
tion to optimum average costs. If the firms in an industry are so few that 
they would encounter serious diseconomies of scale in supplying the 
limit output, they may be unable profitably to forestall entry. But when 
the number of firms becomes such as to allow production of the limit 
output at near-optimum average costs, excessive entry may be profitably 
forestalled by limit-pricing policies and an economical adjustment of 
capacity to demand perpetuated. 16 

The preceding hypotheses are developed on the basis of certain cru- 
cial assumptions, of which the only really controversial one is that po- 
tential entrants to concentrated industries may be significantly influenced 
in their entry decisions by the prices set by established sellers. This 
assumption may or may not find extensive empirical support. But the 
observed price policies in a considerable number of oligopolistic indus- 
tries with apparently effective collusion on price are consistent with hy- 
potheses developed from the assumption, and we may have a thesis of 
real explanatory value. 

The model discussed above also rests on the assumption that the es- 
tablished seller (s) are uncertain of the rivalry which will exist if new 
firms enter the industry. Its conclusions are not essentially modified, 
however, if we assume instead that the established seller or collusive sell- 
ers contemplate effective collusion with any new entrants. If this is as- 
sumed, together with the assumption of a determinate industry demand 
curve and an anticipated lump of entry above a given limit price, the 
analysis develops as follows. The long-run industry demand curve is 
DAD' (Figure 2) and the marginal revenue drawn thereto is Dam, as 

15 E. H. Chamberlin, The Theory of Monopolistic Competition, 1st ed., pp. 100- 

16 It is of course evident that if once an oligopolistic industry gets an excess 
number of firms — whether because the threat of entry is overlooked, or because the 
limit price fails to forestall entry, or because industry demand declines — then there 
is no evident force which will eliminate firms and give such a good adjustment. 
But it nevertheless holds that excessive entry may be deliberately forestalled in the 
manner described. 



before. Similarly the limit price is Q a A; above it the demand for the 
output of the established seller (s) is not DA, and the marginal revenue 
segment Da cannot be exploited by the established seller (s). If the es- 
tablished seller (s) go above the price Q a A or below the output OQ a , 
however, the resultant entry does not render the demand for their out- 
puts indeterminate, since effective collusion with any entrant is con- 

Figure 2 

templated. Presuming that the market share going to the entrant under 
such collusion is calculated, the established seller (s) anticipate a de- 
terminate loss of the total market volume to the entrant at each possible 
price. Then if they raise or attempt to raise long-run price above the 
limit, the anticipated long-run demand for their output becomes some 
determinate curve DD", lying to the left of the industry demand curve 
by distances representing the share of the industry output an entrant 
would obtain at various prices. The corresponding marginal revenue is 

As an approximation we may say that the established sellers' demand 
function is discontinuous horizontally, made up of the segment AD' 
below the price Q a A and the segment A'D above that price. This is not 
precisely accurate, however, since the attempt by established sellers to 
charge more than Q a A, involving attraction of new entry, might result 


in a collusive equilibrium price after entry which lay below Q a A. The 
curve DD" thus can have a meaningful price range which overlaps that 
of AD'. The diagram may therefore be read as follows: the established 
sellers can in the long run sell any of the price-output combinations on 
the line AD' (the quantities OQ a or larger) without attracting entry. If 
they go above the price Q a A, or attempt to, they thereafter in the long 
run can sell any of the combinations on the line DD", with the remainder 
of the industry demand going to a new entrant or entrants. In effect, the 
established seller (s) are given the choice between the truncated industry 
demand curve AD f for exclusive exploitation and the nontruncated 
share of the industry demand curve, DD", for exploitation via collusive 
agreement with the new entrant. 17 The AD' and DD" curves adequately 
compare the revenue alternatives of the established seller (s) with and 
without entry, so long as we neglect any transitional extra profit which 
the established seller (s) might enjoy while raising above Q a A but before 
entry became effective. If such a transitional profit is significant, it must 
be considered as augmenting that long-run average profit which is ob- 
tainable by exploiting DD" after entry. 

If we place the long-run marginal cost of the established seller (s), as 
previously defined, against this demand complex, conclusions generally 
consistent with those already developed emerge. Suppose that the mar- 
ginal cost of the established seller (s) lies at AfCi, so as to intersect indus- 
try marginal revenue in the range am, where price can be below Q a A. 
Provided that total profit at the output OQi, determined by the intersec- 
tion of MCi and am, is positive and exceeds profit at the intersection of 
AfCi and Dm', the established seller (s) will produce OQi, charge a cor- 
responding monopoly price below Q a A, and maximize industry profits 
without attracting further entry. 

Suppose instead that the marginal cost lies at MC 2 , so as to lie above 
industry marginal revenue at the limit output and to intersect the mar- 
ginal-revenue-after entry, Dm', at an output, OQ 2 , which gives a price 
above Q a A (average cost being less than Q a A at OQ a ). The established 
firm(s) can supply the entire industry demand OQ a at the price Q a A, 
or part of the industry demand, OQ 2 , at a higher price. They will choose 
between these discrete alternatives by comparing the lump increment 
to total revenue with the lump increment to total cost which is in- 
curred in moving from OQ 2 to OQ a . 18 If the revenue increment ex- 
ceeds the cost increment (the area under MC 2 between Q 2 and Q a ), they 
will produce at OQ a , sell at Q a A, and exclude entry. The profits of the 

17 The latter curve might also be truncated at a higher level due to an additional 
threat of entry, but we will focus attention here on a single increment of entry. 

18 This lump increment to revenue if the limit output is chosen (as aggregated 
over all future periods) must be considered as reduced by any transitional extra 
profit receivable after raising price above the limit but before entry becomes 


industry will not be maximized, but those of the established seller (s) will 
be. This holds for MC2 in successively higher positions until the cost in- 
crement exceeds the revenue increment between Q 2 and Q a \ even pro- 
duction at OQ a with marginal cost above price is possible. When the 
lump cost increment exceeds the lump revenue increment, price will be 
raised above Q a A and entry will be attracted. (Average cost must of 
course always be covered at the best output or exit from the industry 
will occur in the long run.) 19 

On the supposition of collusion after entry, we thus arrive at con- 
clusions regarding price similar to those developed when assuming un- 
certain rivalry after entry. It may be objected, of course, that if estab- 
lished sellers assume collusion after entry, and potential entrants assume 
it too, then these potential entrants should not be much influenced by 
the current prices of established sellers, and that a limit-price analysis is 
thus implausible. It becomes plausible evidently only if potential en- 
trants are quite uncertain about industry demand and about how they 
will be welcomed by established sellers. But such an incongruity of at- 
titudes is itself not implausible, and the model just developed may thus 
constitute a realistic variant of our first model. 

In summary, a considerable elaboration of the theory of monopoly and 
collusive oligopoly price may be implied if we assume that potential 
entrants to an industry are influenced by the going prices therein, and 
that established sellers anticipate and, if it is profitable, forestall entry. 
Assuming correct appraisal of limit prices by established sellers, we get 
three major possibilities: (1) pricing to maximize industry profit with 
no entry resulting; (2) pricing to forestall entry with industry profit not 
maximized but the profit of established sellers maximized; and (3) pric- 
ing to maximize industry profit but with resulting attraction of addi- 
tional entry. The first two cases find industries already in long-run 
equilibrium; the third finds industries in process of dynamic change in 



The limit-price models just developed, tracing the effects of a sort of 
oligopolistic interdependence between firms already in a concentrated 
industry and potential entrant firms, are essentially variations on the 
general theory of oligopoly price. There is an apparent similarity of 
the construction to the familiar kinked demand curve analysis, but in 

19 The general conclusions also hold if the marginal cost lies so as to intersect 
Dnf where price would be below Q a A, and to lie above industry marginal revenue 
at OQ a . The limit output will be produced and entry excluded so long as the 
revenue increment exceeds the cost increment between the alternative outputs. In 
this case we have the situation that collusive price after entry would lie below the 
limit price which should presumably attract entry. If we remember that the limit 
price is not necessarily the minimum which entrants expect after entry, however, 
this situation is not necessarily anomalous. 


the present case the average revenue as well as the marginal revenue is 
discontinuous, and it is a revenue function for all firms already in an 
industry rather than that for a single oligopolist which is so affected. 
The limit-price analysis applies to collusive oligopoly behavior, whereas 
the kinked demand curve model refers explicitly to the action of a firm 
in noncollusive oligopoly. And the essential conclusions of the two 
models are of course different. 

Various extensions of "limit-price"' reasoning suggest themselves. Thus 
some firms established within an oligopoly may hold price down for fear 
of "fattening" their smaller rivals sufficiently to encourage their expan- 
sion. Departing from the realm of pure oligopoly, established collusive 
firms might extend selling costs beyond the point of industry profit maxi- 
mization in order to discourage entry, so that the threat of entry could 
cause increased costs rather than reduced prices. The model developed 
above could also be elaborated to take explicit account of dynamic 
changes, of varying time lags involved in the "gestation" of new entry, 
etc. Any extensions along these lines might contribute to a more realistic 
theory of oligopoly price. 

The explanatory value of the limit-price hypotheses of course remains 
to be determined. On a priori grounds they appear to be fruitful, al- 
though alternative explanations of observed "low-price" policies may also 
be valid. Some systematic empirical check of the extent of "limit-price" 
thinking within concentrated industries, and especially by price leaders, 
would seem desirable. Direct verification of the crucial hypotheses from 
ex post statistical results, however, would be difficult. The "limit price" 
in any industry, even if recognized, must change over time in response 
to variations in industry demand, in factor prices, in the availability of 
capital to potential entrants, in the age of the industry, and so forth. The 
single limit price of our static long-run analysis becomes in fact a dy- 
namic variable, and would have to be treated as such. It would be diffi- 
cult to establish in a strictly objective fashion, and knowledge of its 
recognition by sellers or of its magnitude might best be gained through 
interview techniques. It would be somewhat easier, on the other hand, 
to check observed price results for consistency with the hypothesis, 
without relying on the subjective impressions of sellers involved. That 
is, objective calculations of the probable limit price (or time series of 
limit prices) could be made for any industry, and the prices actually 
charged in a supposedly collusive oligopoly could then be compared 
both with such limit prices and with prices calculated to maximize in- 
dustry profits. Wherever behavior consistent with the hypothesis was 
found, direct investigation of policies of price calculation might be indi- 
cated. The emphasis often placed on nonprofit motives, uncertainty, 
irrationality, and oligopolistic rivalry as explanations of low-price policy 
in concentrated industries may be unduly heavy, and the effects of 
threatened entry seem certainly to deserve consideration. 


Product Heterogeneity and 
Public Policy 1 



It has been remarked by Mr. Triffin that "for the historian of eco- 
nomic thought, the most revolutionary feature of monopolistic competi- 
tion theories will probably be the unprecedented pace at which they 
conquered their audience." 1 Interpreting this as he does, mainly in terms 
of the appearance in textbooks for the first time of chapters on oligopoly, 
product differentiation, and selling costs, he may be right. But I must 
again lament the widespread misunderstanding of the subject; so that 
what has "conquered" appears often to be something quite foreign to 
the theory, at least as I understand it. Nowhere is this more true than in 
that part of the whole subject which is taken up in this paper: the re- 
orientation of our ideas as to public policy in view of the fact of product 

Public policy must be presumed to seek in some sense the general 
welfare, and hence in the economic sphere it implies a welfare eco- 
nomics. The supremacy of pure competition with its corollary of prices 
equal to marginal costs as the economic welfare ideal is well known. 
Mr. A. P. Lerner's "Rule" is a quick and familiar reference. 2 What is 
perhaps not so well appreciated is how explicitly monopolistic competi- 
tion has been interpreted as merely indicating the nature of the de- 
partures from the ideal which need to be corrected. Thus, although it 
may have reoriented in some degree our ideas as to how the economic 
system actually works, its impact upon our conception of the model to- 
wards which we would move appears to me to have been virtually nil. 
I say "towards which" in recognition of the fact that pure competition 
is evidently a theoretical concept, and that the practical-minded econo- 
mist is often ready enough to point out that "no one has ever advocated 
that it be established." What we want, to be sure, is some kind of "work- 

* The American Economic Review, Vol. XL (Proceedings of the American Eco- 
nomic Association, 1950), pp. 85-92. Reprinted, with revisions, by courtesy of the 
publisher and the author. 

t Harvard University. 

1 Monopolistic Competition and General Equilibrium Theory, p. 17. 

2 The Economics of Control, p. 64 and passim. 



able" competition. But ordinary (purely) competitive theory remains the 
chief source of our criteria as to what should be done if possible, and of 
the direction in which we should move so far as we can. A striking in- 
stance is the subtitle of this part of the program of these meetings: "Can 
the American economy be made more competitive?" The implication is 
evident that if it can be it should. 

Now if pure competition is the ideal, the direction in which we should 
move is very clear. For it is easy enough to show that the actual econ- 
omy is shot full of monopoly elements, and hence that any move to get 
rid of them or to diminish their importance is in the right direction. The 
main point I want to make is that the welfare ideal itself (as well as the 
description of reality) involves a blend of monopoly and competition 
and is therefore correctly described as one of monopolistic competition. 
If this is true, it is no longer self-evident which way we should move, 
for it is no longer self-evident on which side of the ideal lies the ac- 
tuality for which a policy is sought. It is possible that the economy 
should be made "more competitive"; but it is also quite possible that it 
should be made "more monopolistic" instead. Or perhaps, if there are 
faults to be found with it, it should simply be changed, towards some- 
thing else which again involves both monopoly and competition, with 
the frank admission that, since we cannot measure monopoly and com- 
petition quantitatively, there is no way of comparing the actual with 
the ideal on any yardstick involving these concepts. 

Let us proceed at once to the proposition that monopoly is necessarily 
a part of the welfare norm. In abstract terms it seems to follow very 
directly from the recognition that human beings are individuals, diverse 
in their tastes and desires, and moreover widely dispersed spatially. Inso- 
far as demand has any force as a guide to- production x>ne would expect 
entrepreneurs to appeal to them in diverse way^,and_thus to render the 
output of the economy correspond ingly heter ogeneous, using this term 
in its broadest sense to embrace not only the qualitative aspects of th e 
" produ ct itsel f^ but also the conditions su rroundi ng its sale, including 
spatial location. 3 A nd since what people want — an elaborate system of 
consumers' preferences — is the starting point in welfare economics, their 
wants for a heterogeneous product would seem to be as fundamental as 
anythin g could be. Het erogeneity as between producers is synonymous 
with the presence of monopoly; therefore monopoly is necessarily a 
part of the welfare ideal. 

It must be emphasized that any and all monopoly is included within 
the general concept of heterogeneity or differentiation (although there 
is no implication of an identity between the actual and the ideal). A 

3 Apart from the influence of demand, output will also be heterogeneous because 
of the diversity of nature on the side of production; as illustrated by human 
services, both directly and as reflected in the products they create; and by the fact 
that sellers are separated spatially. 


monopoly is simply a product under a single control and significantly 
different from others on the infinite chain of substitutes. This holds 
equally for a patent, a cement producer separated in space from others, 
a local gas utility, a toll bridge, or the A & P. And they are, of course, 
all without exception engaged in competition with others near by on the 
chain of substitutes and with others generally in the system. "Industry" 
or "commodity" boundaries are a snare and a delusion — in the highest 
degree arbitrarily drawn and, wherever drawn, establishing at once 
wholly false implications both as to competition of substitutes within 
their limits, which supposedly stops at their borders, and as to the possi- 
bility of ruling on the presence or absence of oligopolistic forces by the 
simple device of counting the number of producers included. As for the 
conventional categories of industries, it seems increasingly evident to me 
that they have their origin, not primarily in substitution at all, but in 
similarity of raw materials or other inputs or of technical methods used. 
Glass, leather goods, drugs, and medicines are obvious examples. Apart 
from the wide diversity of products embraced by almost any so-called 
"industry," spatial separation of producers within it is an added prime 
obstacle to substitution in most cases. But the main point is that, even if 
lines were arbitrarily to be drawn, they would have literally nothing to 
do with the extent and character of the heterogeneity, either within such 
an industry or beyond it, which would be defensible from the point of 
view of welfare or of public policy. 

All this is in striking contrast with prevailing notions of the signifi- 
cance of product heterogeneity for public policy. The reason is, I be- 
lieve, mainly a difference in the implications of monopolistic competi- 
tion on the one hand and of imperfect competition on the other; and 
the fact that the prevailing notions on public policy have been derived 
largely from an interpretation which follows the latter. It is worth not- 
ing that the terms product and market are used consistently in Monopo- 
listic Competition, not in their usual broad sense, but with reference only 
to the individual firm. There are no "commodities," such as shoes, sheets, 
or shaving brushes, but only groupings of individual products. The term 
"industry" was carefully avoided, and does not appear at all (except 
where its limitations are being pointed out) . By contrast, Imperfect Com- 
petition followed the tradition of competitive theory, not only in identi- 
fying a commodity (albeit elastically defined) with an industry, but in 
expressly assuming such a commodity to be homogeneous. 4 Such a theory 
involves no break whatever with the competitive tradition. The very 
terminology of "imperfect competition" is heavy with implications that 
the objective is to move towards "perfection." 

Even within the terminology of monopolistic competition, the same 
tendencies have appeared in the connotation which the term "differenti- 

4 J. Robinson, The Economics of Imperfect Competition, p. 17. 


ation" has taken on to many as of something superficial. (Hence the 
term "heterogeneous" in this paper.) It is often conceived as describing 
the reprehensible creation by businessmen of purely factitious differ- 
ences between products which are by nature fundamentally uniform. In 
this vein, some have even gone so far as to attribute differentiation, and 
monopolistic competition generally, to "imperfect knowledge," 5 as 
though the individuality of particular products could be dismissed as an 
optical illusion based upon ignorance — a purely psychic phenomenon. 
There seems, on the contrary, to be as much reason for people to lack 
knowledge of the differences between products as for them to lack 
knowledge of their similarities; and there is a good prima facie case for 
believing that "perfect knowledge" (while causing major shifts in in- 
dividual preferences) would leave a system in which there were more 
and stronger preferences than ever. Certainly the consumer research 
organizations, which are engaged in perfecting the knowledge of their 
subscribers as to the goods they contemplate purchasing, are as much 
concerned with differences as with similarities. 

Another device for leveling off the heterogeneous output of the econ- 
omy into a series of purely competitive industries is the distinction be- 
tween "rational" and "irrational" preferences, with its heavy implication 
that a substantial part of actual preferences are of the latter category. 
The distinction is not without its complications; but the test is sup- 
posedly simple: "If a consumer were forced to have B's goods instead 
of A's goods, would he feel worse off after the change had taken place? 
If, in fact, he would consider himself to be worse off, the buyers' pref- 
erence is rational; if not, it is irrational." 6 The conclusion is, of course, 
that if irrational buyers' preference exists, "then the community clearly 
gains by the concentration of the industry's output on a smaller num- 
ber of firms." It need only be commented that the argument, for what- 
ever validity it may have, should not be limited in its application to an 
arbitrarily defined industry, but should be applied generally. On the one 
hand, it may be said that if Palmolive were abolished, people might be 
no "worse off" after they had got used to using Lux and Lifebuoy in- 
stead. But on the other hand, it is equally true that if baseball were 
abolished and bull fights substituted, people might be equally well or 
better satisfied after they were adjusted to the change, in which case 
their preferences for baseball should be classified as irrational. Similarly, 
many people have stopped smoking and, after they got used to it, were 
no more unhappy than before. There is a case, of course, for improving 
knowledge in all these matters, but no reason to think that improved 

5 F. H. Knight, American Economic Review, May, 1946, p. 104; and G. J. 
Stigler, Theory of Price (1946), pp. 214-15, 329, and passim. 

6 Meade, Economic Analysis and Policy, p. 155. I believe the distinction was 
first made by R. F. Kahn in "Some Notes on Ideal Output," Economic Journal, 
March, 1935, pp. 25-26. It is criticized by J. K. Galbraith, ibid., June, 1938, p. 336. 


knowledge would leave us with fewer or weaker preferences. In some 
cases it seems clear that increased standardization of certain products by 
public authority is indicated, as when oligopolistic forces are supporting 
an unduly large number of producers, 7 or when the gain in efficiency is 
judged by proper authorities to be more important than the losses in 
consumers' surplus through abandoning certain products. But the label- 
ing of most preferences within an arbitrarily defined industry as irra- 
tional seems to me to indicate mainly a preference for the purely com- 
petitive ideal, and an attempt, perhaps largely unconscious, to salvage 
it. The alternative is not necessarily to assume that all preferences are 
rational, but only that they are on the same footing — in other words, to 
make no invidious distinctions between them as to rationality on the 
basis of the relative proximity of substitutes. 

It might be added that no invidious distinctions are indicated on 
the basis of whether or not the demands for particular products are in- 
fluenced by selling expenditures. Here again, stress on irrational prefer- 
ences makes an easy transition to the labeling of those established by 
advertising as irrational, and to the conventional sweeping condemnation 
of advertising as a "competitive waste." 8 Granted that the techniques of 
modern advertising are often a shocking affront to good taste, or objec- 
tionable on other grounds, it remains true, so far as I can see, that the 
question of whether advertising is wasteful or not, in the sense of being 
a misallocation of resources, simply cannot be answered by any criteria 
derived from market demand and cost curves — or from indifference 
curves either. Here is a major aspect of "welfare" which appears to lie 
quite outside the conventional analysis of the subject. The general con- 
demnation of advertising as a waste surely has its primary explanation in 
the irrelevancy that it could not exist under the perfectly competitive 

The fact that equilibrium for the firm when products are heterogene- 
ous normally takes place under conditions of falling average costs of 
production has generally been regarded as a departure from ideal con- 
ditions, these latter being associated with the minimum point on the 
curve; and various corrective measures have been proposed. However, 
if heterogeneity is part of the welfare ideal, there is no prima facie case 
for doing anything at all. It is true that the same total resources (either 
within some arbitrarily defined industry or within the whole economy) 
may be made to yield more units of product by being concentrated on 
fewer firms. The issue might be put as efficiency versus diversity — more 
of either one means less of the other. But unless it can be shown that 
the loss of satisfaction from a more standardized product (again, either 
within an "industry" or for the economy as a whole) is less than the 

1 Monopolistic Competition, pp. 100-109. 

8 Meade, op. cit. pp. 165-66 (Meade and Hitch, pp. 176-77) 


gain through producing more units, there is no "waste" at all, even 
though every firm is producing to the left of its minimum point. 

How are the two to be compared — a larger, less heterogeneous output 
as against a smaller, more heterogeneous one? The price system, es- 
pecially in view of its all-pervasive oligopolistic forces and the omni- 
presence of selling costs whose welfare status is uncertain, appears to 
afford no test. If we may allow the individual producer his optimum 
selling expenditure, included as a lump sum in his fixed costs, and con- 
ceive a system in which every producer determines the equilibrium of 
his firm with reference to a demand curve which measures demand for 
his product at different prices while all other prices, products, and selling 
costs do not change, we have in the elasticity of demand one index of 
the strength of buyers' preferences for each product. 9 If adjustment of 
prices along demand curves of this type could be enforced, many firms 
whose profits (perhaps nominal) are protected by the absence of vigor- 
ous price competition 10 would certainly be involved in losses and would 
be obliged to go out of business before a general equilibrium for the 
whole economy were realized. There would be less heterogeneity than 
we find at present, and it would seem that something like what I have 
described elsewhere as a "sort of ideal" 11 would be established. 

Another approach to the same problem is to test old products in- 
dividually for survival and new products for admission to the economy 
by a consideration of the surpluses of satisfaction over cost which are 
sacrificed in one place and generated in another by the transfer of re- 
sources involved. Much of what has been written in this connection 12 
seems to me to be vitiated by entanglement with the standard theory of 
"exploitation" which has evolved out of "imperfect" competition and 
which I have elsewhere 13 shown to be fallacious — a theory in which hired 
factors are held to be exploited by entrepreneurs. But the theoretical 
criterion involved can be adapted to an analysis from which this objec- 
tionable feature is absent. Of course the old bogey of interpersonal com- 
parisons appears at once; also the familiar problem of subsidy to the ex- 
panded firms which, if they had no extra profits before, are now, at the 
lower prices necessary to sell the larger output, losing money. Unfor- 
tunately the matter is too complex to be developed in this short paper. 
Let us only observe that, for whatever it may be worth, the final welfare 

9 The curves of Mrs. Robinson's Imperfect Competition cannot be used for this 
purpose because they are defined as including oligopolistic reactions. Cf. Imperfect 
Competition, p. 21. 

10 Cf. Monopolistic Competition, pp. 100-109. 

11 Ibid. p. 94. With allowance made for the "diversity of conditions surrounding 
each producer" (pp. 110-13) the ideal would evidently involve diverse outputs and 
prices for the individual producers in the system. 

12 Cf. especially R. F. Kahn, op. cit., and J. E. Meade (also Meade and Hitch), 
op. cit., Part II, chap. vi. 

13 Monopolistic Competition (5th or later ed.), pp. 182-84, 215-18. 


equilibrium which emerges from this analysis, as from the preceding 
one, would inevitably involve product heterogeneity; and that it would 
be characterized neither by the equation of price and marginal cost nor 
by production at minimum average cost for the firms involved. Indeed, 
by this procedure, the adjustment required from any starting point might 
as easily be to increase the supposedly excess number of firms as to 
diminish it. As an indication of what is involved, one might under this 
principle even revive that good old newspaper, the Boston Transcript, 
under public subsidy, since many "proper Bostonians" were strangely 
attached to it and no doubt lost heavily in consumers' surplus when it 
finally folded up. 

Let us leave this question of how many products there should be, or 
of diversity, to say a word about the other major type of adjustment 
which has been analyzed in relation to product heterogeneity and wel- 
fare — that of the distribution of resources among a given number of 
products or among a given number of industries. 

It has been proposed that resources be transferred from purely com- 
petitive industries, where price equals marginal cost, to "imperfectly 
competitive industries," where price is greater than marginal cost, and 
similarly from less imperfectly competitive to more imperfectly com- 
petitive industries, until the ratio of price to marginal cost is the same 
everywhere. Such a proposal may be dismissed at once on two grounds, 
either one of which alone is sufficient: ( 1 ) the boundaries of an industry 
being arbitrary, it is quite meaningless and (2) the demand and cost 
curves of different firms within any industry are highly diverse as to 
elasticity and shape. For these two reasons we must abandon altogether 
the idea of transferring resources in some vague way to an industry, and 
face the question of the firms to which they are to be attached. 14 

What, then, of equalizing the price-marginal cost ratio as between 
firms in the economy? Apart from other difficulties, I believe there is a 
fatal objection to such a conception; viz. the generally prevalent oligop- 
olistic relationships between firms. The logic by which this proposition 
is usually developed envisages each firm as an isolated monopoly, iso- 
lated in the sense that its output and price may be adjusted without ap- 
preciable effect on any other single firm. But where oligopolistic influ- 
ences are present, there are two points to be made. First, the demand 
curve for any one firm, which would indicate the effect on its price of 
adding resources to it, cannot be known without knowing which of the 
many possible patterns of behavior under oligopoly will govern the case 
at hand. In fact, since adding resources to any one firm would, by lower- 
ing its price, inevitably shift the demand curves of others economically 
near it (since every curve is drawn on the assumption of given prices for 

14 Mr. Kahn's analysis explicitly assumes industries in which competition is "uni- 
formly imperfect" (op. cit. p. 21 n.), and thus lays down principles for a wholly 
imaginary problem. 


other firms), there seems to be no escape from abandoning the concep- 
tion of transfers between firms considered to be independent, and recon- 
ceiving it in terms of groups of some kind. Second, the effect on welfare 
of adding resources to one firm, where oligopolistic interdependence is 
involved, is a function of whether or not, and in what quantity, resources 
are being added at the same time to others economically near. Even as- 
suming that the price behavior could be directed according to some so- 
cially enforced rule, the major problem would still remain of finding the 
rule in welfare terms. I very much fear that, because of oligopolistic 
interrelationships between the welfare contributions of firms, we are re- 
duced to asserting merely that resources should be transferred from one 
place to another in the system whenever the net effect will be to increase 
welfare. This is not very illuminating. 

In conclusion, the consequences of product heterogeneity for welfare 
economics have been either ignored or seriously misunderstood. Monop- 
oly elements are built into the economic system and the ideal necessarily 
involves them. Thus wherever there is a demand for diversity of prod- 
uct, pure competition turns out to be not the ideal but a departure from 
it. Marginal cost pricing no longer holds as a principle of welfare eco- 
nomics (not even for toll bridges) ; nor is the minimum point on the cost 
curve for the firm to be associated with the ideal. Selling costs may no 
longer be excluded from the problem or dismissed as an obvious waste; 
yet the impossibility of discovering from the standard welfare techniques 
what is the socially ideal expenditure on selling suggests that the tech- 
niques are unduly narrow. It has been impossible to discuss in this paper 
whole families of new problems which put in their appearance with a 
recognition of the fact that products themselves are variables and that 
there must be norms for them as well as for prices, costs, and outputs. 
What has been called the "new welfare economics," instead of being on 
a "secure basis" as suggested by Professor Hicks, 15 has quite miscon- 
ceived a whole set of major problems. It is badly in need of a general 

15 "The Foundations of Welfare Economics," Economic Journal, December, 1939, 
p. 711. 


Competition: Static Models and 
Dynamic Aspects* 

By J. M. CLARK' 

This paper is an attempt at a contribution to the endless effort to re- 
duce or bridge the gap between theory and reality; and in particular the 
gap between theories of cost-price equilibrium and theories of produc- 
tion, growth, and development. 1 Theory must always be the department 
of oversimplification; but dynamic simplifications are at least different 
from static. They include the tendencies toward equilibrium. But because 
these tendencies never reach their static limits, dynamic theory cannot 
use any features which are needed only to enable a model to attain this 
impossibly precise completeness — especially if these features are incom- 
patible with the conditions of progress. 

It follows that fully dynamic theory is bound to lack certain character- 
istics which are, to many theorists, the essential earmarks of theory. It 
must accept indeterminateness, with some margin of individual discretion 
in business action, and uncertainties of different kinds from those thev 
are accustomed to dispose of by assuming a premium for uninsurable 
risk. Some of these uncertainties are positive aids to effective — not per- 
fect — competition. If such theory is to be accepted as theory, it seems 
that the first step which theorists can contribute is the setting up of an 
appropriate conceptual framework, including appropriate tool-concepts. 
Even within theoretical writings, many or most of the pieces of this 
framework exist as scattered fragments; but the writer is not aware that 
the framework has been assembled and articulated. The present paper 

* The American Economic Review, Vol. XLV (Proceedings of the American Eco- 
nomic Association, 1955), pp. 450-62. Reprinted, with revisions, by courtesy of the 
publisher and the author. 

t Columbia University. 

1 The remarkable group of papers presented at last year's [1953] meetings of this 
Association and the proceedings of the 1951 conference of the International Eco- 
nomic Association (Monopoly and Competition and Their Reflation, E. H. Cham- 
berlin, editor [London, 19541) include an appeal for a dynamic theory and confirm 
the existence of an impressive body of factual material large enough to afford a basis 
for beginnings at fresh generalizations. They present a challenge to which theory 
should attempt to respond. 



is a modest attempt to bring together enough pieces to afford some sug- 
gestions of what the structure might look like. 


In the theory of a generation ago, competition played a twofold role: 
as an agency to eliminate excessive and exploitative profits and as a stim- 
ulus to technical progress — a keener stimulus than monopoly would af- 
ford. As a stimulus to progress, competition included both the carrot 
and the stick (the carrot of profits to the successful innovator and the 
stick of losses for those who fall behind) and elimination for those who 
fall behind substantially and persistently. Temporary monopoly, under 
the patent system, was also recognized as a force for technical progress. 
Later, Schumpeter's theory of innovation made its mark, but without 
being integrated into general competitive theory. 

Subsequent developments, leading to the body of competitive theory 
now prevalent, have concentrated on more precise definition of the con- 
ditions necessary to complete or "perfect" cost-price equilibrium — con- 
ditions that turn out to be nonexistent. The model then becomes an ana- 
lytical device and, as such, a step toward explaining why things do not 
behave like the model, the next step being to take account of the inevi- 
table departures and the conditions accounting for them. So far, this 
process has produced certain other models, still highly simplified, of mo- 
nopolistic competition and oligopoly, with and without product differ- 
entiation. The preoccupation of these models with cost-price equilibrium 
is obvious. What may be less obvious is that static limitations are built 
into their underlying tools of analysis — demand curves and cost curves 
— which therefore need modification for dynamic purposes. 

This prevalent body of theory is not a theory of production, as 
growth theory must necessarily be. Insofar as it deals with economy and 
efficiency of production, it does so in terms of optimum or less-than- 
optimum scale of production (size of plant or firm) on a fixed long-run 
cost curve; that is, on the obviously static assumption that the "state of 
the arts" remains constant. 

A more troublesome consequence of the model consists of the nor- 
mative conclusions that are, rightly or wrongly, drawn from it. As a 
standard of so-called "perfection," it is one-legged, focusing on the es- 
sentially static objective of cost-price equilibrium, to the neglect of the 
dynamic objectives of progress. This one-legged standard is frequently 
treated, without warrant, as an ideal. And because it is unattainable, all 
actual or possible conditions are, by comparison, judged inferior, mo- 
nopoloidal, or actually monopolistic. 2 This interpretation Chamberlin 

2 The opposite case— of pure but imperfect competition (e.g., agriculture) 
which tends to be unduly severe — is commonly neglected. 


himself has flatly repudiated, stating that perfect competition is not the 
ideal, and the ideal includes elements of monopoly — which might mean 
no more than such elements of uniqueness as most businesses inevitably 

The idea of the omnipresence of harmful departures from perfect 
competition seems to underlie the challenge of Professor Galbraith, who 
says in effect: according to theory, our economy ought to be suffering 
all the evils ascribed to monopoly; why isn't it? Galbraith suggests that 
it may be the theory that is wrong rather than the facts of the economy, 
and this contains much truth. But when he suggests that the trouble with 
the theory is its adherence to competition and recommends throwing 
competition out the window in favor of a subsitute, he is surely over- 
playing his hand. I would contend that the trouble with the theory is not 
its adherence to competition, but its too formalized conception of what 
competition is. This includes, first, the one-legged character of its 
standard of perfection, and, second, the fact that its picture of reality 
is highlighted by models of oligopoly and monopolistic competition 
which do not tell the whole story about the cases with which they deal, 
and do not do justice to the competitive forces that exist, including forces 
of progress and forces working toward serviceable — not perfect — cost- 
price adjustments. The cases include monopolistic evils, but they also in- 
clude — more typically, I suspect — cases in which the reality behaves bet- 
ter than the models, including the model of perfect competition, which 
sacrifices important factors of dynamic progress. 

But recognition of this lies under a heavy semantic handicap, if it has 
to take the shape of stating that the ideal includes "elements of mo- 
nopoly." This raises questions, not as to the facts, but as to the expedi- 
ences of the terminology. It seems unfortunate if economists who defend 
realistic forms of competition thereby expose themselves to the charge 
of defending monopoly. This danger is not imaginary. The outstanding 
case is that of product differentiation — a "monopolistic" element which 
has in itself no necessary antitrust significance. I have myself said, of 
product differentiation (Economics of Overhead Costs, page 418): "In 
a sense each competitor has a monopoly of the difference in quality . . . 
and this qualified monopoly is a feature of the typical 'competitive' mar- 
ket." I would not say he had a monopoly of his product, but only of the 
quality differential; and I would not suggest this as standard general- 
purpose terminology, to be used, among other things, in antitrust 
connections. With such uses in view, it would appear helpful if, in our 
general-purpose terminology, "monopoly" began with obstructions to 
imitation, and "incipient monopolization" began with actions of an ob- 
structive character. 

There is food for thought in the discrepancies or contradictions be- 
tween pure theory and practical or policy judgments. For example, the- 
ory appears to regard product differentiation as always a shelter from the 


rigors of competition; but the Interstate Commerce Commission has ap- 
parently found that the element of differentiation between rail and truck 
transport tends to cause their competition to assume a cutthroat charac- 
ter. And businessmen recognize that a rival's product innovation can be 
a very aggressive and formidable method of competition. A manufac- 
turer who has no competitors at his immediate location is not for that 
reason a monopolist under the antitrust laws; but he must watch his step 
as to what he does with the "element of monopoly" resulting from his 
unique location, under the Staley and other basing point cases. On dis- 
crimination and freight absorption, neither theory nor antitrust policy is 
clear and unambiguous. Excess capacity tends to be treated by theory 
as a monopolistic symptom, 3 but in antitrust cases it may or may not be 
so treated, depending on circumstances. And businessmen tend to the 
view that competition does not begin until they have difficulty disposing 
of capacity output, and becomes more severe as excess capacity gets 

The perfect competitive market of theory is one in which prices are 
identical and each producer knows the others' prices and profits, which 
implies knowing their costs. In practice, too perfect identity of prices 
is legally suspect, and so is interchange of information on prices, as in 
open price associations, or information on costs, or even the use of a uni- 
form cost accounting system. All these cases serve to suggest things 
which current equilibrium models neglect and which dynamic theory 
may need to recognize. 


In the light of the things we want competition to do for us, what are 
the features it needs to have which are implied in these objectives? I 
will put first the elements required for progress as being most important, 
since even a small continuing gain outweighs a substantial gain of the 
once-for-all variety. 

First comes progress in economical methods of production. Under 
competition, this implies that some take the lead and others follow, while 
managements are changed or firms are eliminated if they fall too far be- 
hind. But an excessive rate of elimination of firms may be unhealthy for 
an industry. To avoid this without retarding progress, the bulk of the 
followers needs to be able to keep near enough to the leaders to stay in 
the race. 

Secondly, we want competition to afford customers an amply differ- 
entiated range of qualities and types of any given product to choose 
from. This is, of course, wanted for its own sake; but in terms of prog- 
ress it means that the producers are exploring — and influencing — the cus- 

3 See, especially, Chamberlin, Theory of Monopolistic Competition (5th ed.), 
pp. 109, 171. 


tomers' preferences and potential preferences, and products evolve in the 
directions these preferences indicate. Whether they evolve in the best 
directions is, of course, a different question. Product differentiation costs 
something, mainly because of the selling effort that is bound up with it. 
(For that matter, maintaining homogeneity of product costs something, 
too. And homogeneity of each brand is necessary for effective differen- 
tiation between brands.) The combination of selling costs with research 
and testing introduces an element of overhead cost which makes for large 
size and limited numbers of producers. But in a market as large as the 
American, this seldom goes to lengths that spell natural monopoly. 

Thirdly, we want new products developed; and this is a necessary 
correlative of more efficient methods of production, if an increase of 2 
per cent a year in productivity is to be absorbed without technological 
unemployment. New products, even if patented, are generally exposed 
to imitation and competition after a delay which is, by historical stand- 
ards, not long. 

A second group of objectives is concerned with the diffusion of the 
benefits of progress, to customers in lower prices or to those who con- 
tribute factors of production — chiefly workers — in higher real rewards. 
This involves a progressive sharing of the rewards of successful innova- 
tion, and the sharing becomes complete when the improvement has be- 
come part of the generally available state of the arts, from which special 
profits can no longer be made. This diffusion means that what Veblen 
called "the wisdom of the ancients" is not "cornered" by current busi- 
ness enterprise; only the recent advances made in it, and these only par- 
tially. The diffusion is wanted for its own sake, but it also has a special 
role in the incentives to progress. What it means is that a renewal of dif- 
ferential profits can be had only by renewed innovation. Thus it is part 
of the system that can keep incentives to continued innovation alive in 
full strength. This role would be vitiated if the diffusion were instanta- 
neous and complete, since then innovation would bring no rewards to 
the innovator. The dynamic system is not one of elimination of profits, 
but one of erosion and re-creation, both of which are jointly essential. 

For the economy as a whole, this process implies the creation, reduc- 
tion, and re-creation of differential rewards in different industries, as 
well as for different firms. Professor Harberger's paper at the 1953 
[American Economic Association] meeting developed a case for the the- 
sis that existing differentials are relatively unimportant defects, by con- 
ventional utilitarian standards. This argument may not prove that we 
should be unconcerned about them; but it does strongly suggest that 
their dynamic impact is more important than their static. The goal is to 
subject them to active wearing down and prevent them from becoming 
so enduringly entrenched that they need not be earned and re-earned by 

The diffusion process also affects growth via the distribution of in- 


comes, which in turn affects the market for the increased flow of goods, 
new and old, which progress makes available. This has two main dimen- 
sions: inequality between persons and between different industries or 
sectors of the economy. If one sector of the economy is progressive, it 
will be handicapped in realizing its potentialities if most of the other 
sectors are too poor to buy much of the products of the dynamic sector 
and too stagnant to increase their purchases as the productive power of 
the dynamic sector increases. 4 Development needs to be not too badly 
unbalanced; and from this standpoint well-diffused competition is favor- 
able. Something similar is true of personal distribution, a fair approach 
to equality being favorable to the kind of demand on which mass produc- 
tion depends. On the other hand, new products have often been intro- 
duced first as luxury items, appealing to the rich or well-to-do, and later 
coming within the reach of mass demand, as increased sales and the re- 
sulting productive economies reinforced one another cumulatively. 
Nowadays, with the American mass market thoroughly established, it is 
sometimes possible to promote a new product for mass volume from the 

What is not obvious about all this is the part played by competition. It 
opposes the inequalities of intrenched privilege; but if the labor market 
were thoroughly competitive, inequalities of wages based on unequal 
performance might be, by and large, greater than they are. On either 
basis, however, this country would have a great mass market plus a more 
limited luxury market, not too far above the mass market and shading 
into it by easy gradations. Competition also implies mobility of produc- 
tive resources, which is needed for adjusting to the inevitable inequali- 
ties in rates of growth in different parts of the economy. And in a grow- 
ing economy, mobility is possible with a minimum of actual shifting of 
persons or capital. 

A phase of this diffusion process is the creating of a market in which 
a customer may have considerable confidence that the offerings of differ- 
ent sellers are of approximately equal value and that he will not be se- 
riously victimized if he fails to shop around, looking for the best offer- 
ings. This (limited) assurance is of real value to the customer who has 
many demands on his or her time; but it works only if enough customers 
do shop around. A competitive market tends toward one price to just 
the extent that customers watch for differentials and take advantage of 
them. This is the customers' essential role in a competitive system. 

A third group of objectives is concerned, not with products and prices, 
but with the conditions of competitive rivalry in itself. It regards free- 
dom and opportunity as ends in themselves. It is concerned with the hu- 
man impact of competitive pressures on those who are exposed to them. 

4 For the suggestion here adopted, I am indebted to R. Nurkse's Proble?ns of 
Capital Formation in Underdeveloped Countries (1953). 


But especially it is concerned that efficiency and the diffusion of its bene- 
fits should not be dependent on the good will and arbitrary decision of 
private producers, nor on direct governmental order, but on a situation 
in which normal business motives impel business units (acting independ- 
ently) to conduct that will tend to further the desired results. This goal, 
like the goal of diffusion of gains, is overrigidifled in the theoretical con- 
cept of an economic law which dictates economic conduct and its results 
with a precision from which there is no escape. 

This would reduce economic freedom to a paradox, since the mean- 
ing of freedom resides in a margin of discretion in choosing one's course 
of action. One of the tasks of the kind of theory here contemplated is 
to define the margins of discretion that producers have and the margins 
they should have or can safely be allowed to have. They may, for exam- 
ple, face uncertainties and be governed by longer or shorter perspectives 
and broader or narrower conceptions of their interests. These all belong 
in a dynamic theory. The essential limits on private discretionary power 
are presumably satisified if the discretion of business units does not ex- 
tend to doing things that obstruct the progress of the "generally avail- 
able state of the arts" or prevent the public from getting the benefit of 
it. This is admittedly a not-too-precise standard; and may need some care 
to avoid tautology, centering in the phrase "generally available." It has 
a logical kinship with Marshall's "representative firm," and I believe it is 
not wholly meaningless. 

Given these goals, how should one use them to judge a system? A gen- 
eral judgment may be influenced by appraisal of the performance of the 
system in the large. But for purposes of specific policy in actual cases, 
rates of progress and fairness of profits are too uncertain, dependent on 
too many irrelevant circumstances; and their use to determine legality 
would hardly be consistent with a free private economy. Instead, judg- 
ment must hinge on whether conditions are of a sort inherently adapted, 
by and large, to promote these ends. Three key tests may be suggested: 
free and independent action, incentive to do the kinds of things that are 
called for, and capacity to do them. 

As to the first, it is not easy to make people compete if they do not 
want to; but collusion can be checked and protection can be afforded to 
the more competitively minded mavericks who will exert pressure on the 
others. Dynamic change also tends, as we shall see, to keep rivalry alive. 
As to incentive, mention has been made of the carrot and the stick, of 
the joint importance of differential gains and their erosion, but no precise 
optimum rate of erosion appears to be definable. Where progress re- 
quires heavy research outlays, atomistic industries are handicapped as to 
both incentive and capacity: as to capacity for obvious reasons and as to 
incentive because the output of the single enterprise is so small that in- 
dividual gains fail to measure the importance of the resulting progress 
for the industry as a whole. Thus in agriculture this element of progress 


depends heavily (though not exclusively) on governmental research and 


A theory of competition as dynamic process must be not a model but 
a framework within which many models may find their places, including 
equilibrium models as limiting hypothetical cases. As I have said else- 
where (Monopoly and Competition and Their Regulation, pages 326— 
28), the process includes initiatory action by a firm, responses by those 
with whom it deals, and responses to these responses by rival firms, to 
which one should add the subsequent rejoinders of the initiators, plus 
any actions that may be taken on a basis of anticipation. (For example, 
a rival may react defensively, before his customers have time to switch 
to a competitor.) The moves and the responses may affect productive 
processes, products, selling efforts or prices, or various combinations. 
They may be aggressive, defensive, or counteroffensive. Fully dynamic 
theory needs to conceive these patterns as themselves subject to change, 
over longer periods of time. Marshall's life history of a firm may be 
matched by life histories of products or product variants, and of the 
marketing patterns connected with them. Stages of exploratory introduc- 
tion, aggressive expansion, defense of established position, and decline — 
all have different features. 

For dynamic theory, a key element is a time interval between moves 
and responses, or a time distribution of responses; and this time factor 
is typically essential to give firms an incentive to make competitive 
moves, by giving them a chance for a temporary gain before their moves 
are neutralized by the defensive or counteroffensive responses of rivals. 5 
Or there may be other elements tending to prevent immediate and com- 
plete neutralization. 

Perhaps the chief common feature of competitive action is an expecta- 
tion of a gain that is sure to be eroded; and it is not convincing to 
make competition hinge on businessmen being so uniformly stupid that 
they experience this repeatedly without learning to anticipate it. At least 
the more intelligent must be thinking in terms of a result that will out- 
last the erosion process; and in this light the decisive motive must be a 
preference for eroded profit on a larger volume of business, over a sim- 
ilarly eroded profit on a smaller volume. I speak of an "eroded" profit 
rather than "zero profit," partly because before profit from one success- 
ful move is reduced to zero, other moves may have brought other profits, 
and partly because "zero profit" carries a misleading precision. It is inter- 
esting that Chamberlin's tangency theorem for differentiated products 

5 After writing the above, I find these time elements recognized by A. Henderson: 
"The Theory of Duopoly," Quarterly Journal of Economics, November, 1954, 
pp. 565, 580. Also by the Civil Aeronautics Board, according to a report of its 
decision in Air Freight Tariff Agreement Case, 14 C.A.B., 424, 428, 430 (1951). 


involves conduct of the opposite sort: pricing for short-term profit, 
which after erosion leaves the producer with a smaller volume than he 
might have had with a more farsighted policy. This can happen; but the 
diagrammatic proof that it must happen wherever product is differenti- 
ated, is unconvincing once one makes adjustments in the concepts of cost 
and cost curves and demand curves which are needed for dynamic the- 

Cost, conceived as including the minimum return necessary to attract 
capital and enterprise, includes, under dynamic conditions, large ele- 
ments of unpredictable obsolescence, thus making "zero profit" an indefi- 
nite, but substantial, quantity. This makes it rational to prefer larger to 
smaller volume, at "zero profit"; and this seems clearly to accord with 
the characteristic American business emphasis on the importance of 

As for the envelope cost curve, so frequently used, each point on it 
must logically represent the most economical method of producing a 
given output, if a plant could be adapted to that output and never have 
to change. It takes no account of fluctuations or of growth by substan- 
tial-sized units or of provision of reserve capacity to be ready to handle 
contingent increases, including those which a vigorous sales-promotion 
campaign might bring, if it proved successful. No competitor likes to 
expose himself to losing customers to his rivals because he cannot fill 
their orders promptly enough. Thus on a more realistic curve of cost 
with scale or size of plant or enterprise, each point would represent, for 
a given size or expected average output, the average cost that might be 
expected during the period in which the plant or enterprise remains of 
this size, allowing for periods when it will be working at part capacity 
and relatively high average cost and other possible times when it may 
be working overtime or otherwise be pushed beyond its optimum rate. 
Such a point would not be on the short-run cost curve but above it, and 
a long-run cost curve connecting such points would intersect the short- 
run cost curve, not be tangent to it. 

The real point is that the whole range of actual or probable movement 
on the short-run curve comes into play, not merely the point of supposed 
tangency. At that point, long-run and short-run marginal cost are equal 
as drawn, thus obscuring the fact, which is crucial for explaining actual 
competitive behavior, that at most times short-run marginal cost is quite 
substantially below long-run marginal cost (which in turn is likely to 
differ little from average cost) while occasional peaks of demand may 
push short-run marginal cost well above both long-run marginal cost and 
average cost. A firm may price well above marginal cost — pricing for 
maximum profit as determined by a sloping individual demand curve — 
and may still fail to cover average cost. 

A timeless two-dimensional demand curve of the conventional sort 
leaves out of account the fact that the effect of a given price, or price 


differential, on the volume of sales is a function, among other things, 
of the length of time during which it has been in effect. The effect may 
reach its limit in time, or may be indefinitely progressive, if the stimulus 
can be maintained. This time dimension has been recognized at times by 
economists — generally when they were not speaking as theorists. It 
means, among other things, that the effect of a given price on sales vol- 
ume depends on the previous price or price situation; and that the curve 
is not fully reversible. 

A given price may take full effect in minutes (as on a produce ex- 
change) or may take decades where it requires changes in consumers' 
ways of life which have strong inertia. An example might be the effect 
of reduced electric rates in stimulating household appliance use of elec- 
tricity. This case is typical in that the active variable is better described 
as a price policy than a price, and acts jointly with promotion of the 
sale of household equipment. Similar comments apply to alterations of 
the product and moves in the area of sales promotion. This complex of 
variables would overload any possible system of graphic presentation. A 
family of three-dimensional surfaces — the third dimension being time — 
with a different surface for each initial price or price situation, would 
still be a simplification. 

Both for cost curves and demand curves, movements along the long- 
run curve involve a shifting of the short-run curve, setting up a new 
one of different dimensions from the former one. In the diagrammatic 
development of Chamberlin's tangency theorem, the treatment of cost 
and of demand in this respect is nonsymmetrical. Long-run movements 
of demand are treated exclusively as unanticipated shifts of the short-run 
demand curve, and producers are assumed to be governed exclusively by 
the short-run curve prevailing at the moment and not by longer-run ex- 
pectations. Thus his diagrams utilize a short-run demand curve, placed 
against a long-run cost curve. A different method could lead to signifi- 
cantly different results. 

The responses of those with whom a competitor deals bring in an eco- 
nomic process which is naturally left offstage by equilibrium theory but 
is essential to dynamics; namely, market canvassing and bargaining. Gal- 
braith has taken advantage of theory's neglect of this process to annex it 
as one sector of his rather heterogeneous category of countervailing 
power, treating it as a substitute for competition which has lapsed into 
passivity. Actually, as theory should have been insisting all along, this 
activity of customers is a necessary complementary part of the process 
of competition, serving to bring to bear and make effective the competi- 
tive alternatives that may exist. Without it, competition would become 
passive, though, as earlier noted, many customers can neglect it without 
serious results if enough others attend to it. 

Three grades of customer activity may be distinguished. First, buyers 
may simply canvass the available alternatives and choose the one they 


prefer. This increases the cross-elasticity of demand on which the effec- 
tiveness of competition depends and tends toward equalizing the attrac- 
tiveness of different sellers' offerings. Second, some buyers may try to 
get a better bargain than is currently offered, using their power to shift 
their patronage as a leverage. It is chiefly the larger buyers who have 
a chance to do this, and it is likely to result in discriminatory favors, 
granted frequently (but not exclusively) by the smaller and less power- 
ful suppliers. Letting contracts on a basis of sealed bids is one variant of 
this process in which one large buyer tries to bring to bear a specially 
active form of competition for his individual business out of which he 
hopes to get a differential price. Finally, there is the buyer with real mo- 
nopolistic power, or bilateral monopoly, in which it seems that almost 
anything might happen. 


The foregoing ideas suggest, among other things, some modifications 
in current theorems, including those of pure oligopoly, and the tangency 
theorem for competition with differentiated products and sloping indi- 
vidual demand curves, tending to reduce the departures of these models 
from older conceptions of cost-price equilibrium. In the case of pure 
oligopoly, the indicated modifications begin with the question: Under 
what conditions is it natural for a firm to expect an increase or decrease 
in its price to be followed or not followed by its rivals, and how may 
these expectations be modified by devices available to a firm, other than 
a simple increase or decrease in its list prices? The suggestion has been 
made that an increase is more likely to be followed if costs have risen 
generally. And this possibility can be sounded out by announcing in ad- 
vance an intended increase, which can be withdrawn if rivals do not fol- 
low suit. If there is much unused capacity, a price advance is unlikely 
to be followed, and existing prices are likely to be subject to downward 
nibbling. But if prices get ruinously low, a move back toward a more 
normal level may be followed though likely to be unstable. 

The number of rivals plays a part in that, if unused capacity exists, 
the actions of the group tend to be controlled by that member whose 
ideas lead to the lowest price policy. Numbers also play a part in that 
the initiator of a price reduction which is not instantly neutralized can 
stand to gain more than any one of his rivals stands to lose. Thus the 
outcome may be affected by differences between the slopes of what 
may be called the aggressive and the defensive demand curves, as well as 
by differences between the curves envisaged by different firms. More 
important, perhaps, are devices for making concessions which will not 
apply to a firm's whole business but mainly — at least for a time — to the 
added business it hopes to gain. When demand is strong enough to uti- 
lize capacity fully, prices would tend to rise under either oligopoly or 
old-fashioned competition; but the rise under oligopoly may be more 


moderate because influenced by a longer time perspective. Another miti- 
gating factor is the existence of stand-by capacity, of only moderately 
inferior efficiency. 

As to the case of differentiated products, if this is treated in terms of 
short-run cost curves, price may be above short-run marginal cost and 
still be below average cost, and below long-run marginal cost. A fairly 
typical hypothesis would be that the bulk of production is on a scale 
sufficient to afford the bulk of the economies of size, implying that the 
slope of the long-run cost curve is sufficiently flat and uncertain to be 
an unimportant factor, except for definitely small firms. I cannot con- 
ceive a firm deliberately following a price policy which, for the sake of 
a small price advantage, would condemn it to be one of these small firms 
when a more aggressively competitive policy might lead to progressive 
expansion. Such a size-limiting policy involves outright danger to sur- 
vival, not merely larger or smaller profits. 

While competition may emphasize model changes and sales promotion, 
these in themselves tend to create uncertainty whether a price reduction 
will be fully and instantly met. Thus they tend to prevent price differ- 
entials from becoming so settled in trade custom that a price reduction 
by one firm is sure to be instantly neutralized by the others. Thus, while 
competition spurs innovation, innovation in turn helps to keep price com- 
petition from settling into an inactive rut. 

In both types of case, needless to say, a workably competitive result 
requires that there be no outright collusion. But it does not necessarily 
require the ignoring of rivals' probable reactions; merely that their ex- 
pected reactions should not be so prompt and complete as to wipe out 
all competitive incentive, either hope of gain or fear of loss from letting 
others "get the jump." This latter may be the most important factor. 


The foregoing is a mere fragment of the whole picture. It by no means 
implies that all industry is satisfactorily competitive. It leaves much room 
for inequalities in the impact of competitive pressures in different parts 
of the economy. The most serious inequalities are those between the 
pressures of competition in industry and trade, on the one hand, and, on 
the other hand, the pressures it would produce, if unmitigated, on agri- 
culture and labor. One explanation for this inequality seems to carry the 
implication that competition in these fields would impose no undue bur- 
dens if only industry and trade were not shot through with monopoly, 
raising the prices of the things workers and farmers buy. Without deny- 
ing that there is a measure of truth in this, I venture to suggest a dif- 
ferent hypothesis as more important: namely, that in industry and trade, 
the producers' side of the market adjustment is dominated by entrepre- 
neurs' dollar expenses, which will (with qualifications) veto production 
if receipts do not cover them, while in the case of labor and one-family 


farming units, the producers' side is a genuine supply schedule which is 
highly inelastic and may be actually backward-sloping — meaning that a 
reduction of wages or farm prices tends to result, if anything, in more 
crops raised, or hours of work offered, rather than less. 

This is, of course, not a new idea, but it seems to carry a two-sided 
consequence. One side is that labor and agriculture might need some 
mitigation of the pressures of competition, even if the rest of the econ- 
omy were unqualifiedly competitive. In other words, if we got rid of all 
monopoly in industry and trade, that might still not make the economy 
safe for unmitigated competition in labor and agriculture. The other side 
is that an acknowledged need on the part of labor and agriculture for 
mitigations of competitive pressures is neither proof nor measure of the 
dominance of stultifying monopoly in the rest of the economy. 

There can be no certainty that competition will remain vigorous in 
American business. The necessary conditions are a fascinating subject 
for speculation. Tentatively, they appear to include three mutually inter- 
acting areas: public policy expressed in the antitrust laws, the aggressive 
psychology and adaptability of the American businessman, and the im- 
pact of continuous change in techniques, products, and channels of trade, 
tending to keep things stirred up and uncertain and to prevent com- 
petition from lapsing into routine passivity. In the troubled years that lie 
ahead, this complex of factors is destined at best to an insecure existence. 
To keep competition healthy requires the traditional eternal vigilance. 
To recognize competition when we see it, in its present-day forms, re- 
quires not only factual study but some reorienting of the traditional 
framework of theoretical concepts in which the facts may find their in- 


Fritz Machlup:! The call for dynamic, more dynamic, and fully 
dynamic theory is repeated over and over again. I fear that the dynamics 
of the meaning of the word "dynamic" is too great for the term to be 
of continued usefulness if we want to communicate and not only to 
orate. Professor Clark wants to communicate, I know, and therefore I take 
the liberty of commenting a bit on his choice of language. 

He speaks of a gap between theories of equilibrium and theories of 
growth and development. As I see it, equilibrium is a tool of thinking 
needed to explain change, no matter whether it is change in market prices 

* The American Economic Review, Vol. XLV (Proceedings of the American Eco- 
nomic Association, 1955), pp. 481-82 and 487-90. Reprinted by courtesy of the pub- 
lisher and the authors. Only those parts of Professor Machlup's discussions that dealt 
with Professor Clark's paper are reprinted here. 

t Johns Hopkins University. 


or in business investment or in employment or in the rate of growth. 
Equilibrium is a valuable concept both in models which disregard time 
lags — static theory — and in models which present sequences in time — 
dynamic theory. Since a theory of growth will ordinarily use the equilib- 
rium concept, there can be no contrast between theories of equilibrium 
and of growth. What Clark meant, I suppose, was the contrast between 
"static theories of price and output" and "theories of growth and devel- 

Price theory itself may be static or dynamic. I once gave a graphic de- 
scription of a "dynamic or time-sequence analysis of effects of a selling 
campaign," in which five different outcomes — different sales volumes at 
one price and different prices for one sales volume — were shown to re- 
sult from a given selling effort, depending on the chosen sequence of ac- 
tions and on the time intervals between the steps. The five points would 
all coincide if sequences and intervals of time did not matter and de- 
mand curves were assumed to be reversible (Economics of Sellers' Com- 
petition, pages 185-89). Whether the complicated dynamic model has to 
be used or the simple static model will do, depends on the kind of prob- 
lem at hand; that is, not on whether sales actually depend on the time 
sequences and intervals but on whether the differences really matter for 
the kind of answer required. Clark gave a very similar description but 
did not say whether he believes that the dynamic model is always su- 
perior. If he thinks so, I would disagree. Dynamic theory is better only 
where it is needed; where the analytic job can be done with the simpler, 
less realistic, static theory, the latter is preferable. 

"Fully dynamic theory," Clark says, "must accept indeterminateness." 
I suppose he means to suggest this as a definition; namely, that a theory 
which is both dynamic and indeterminate should be called "fully dy- 
namic." I do not know whether this would be helpful. Oligopoly theory, 
to be sure, will usually have to be dynamic, will have to make large al- 
lowance for uncertainty in the anticipations of sellers, and will have to 
leave a large scope for indeterminateness. (See my chapter on "Oligop- 
olistic Indeterminacy," op. cit., pages 414 ff.) I should prefer to use lan- 
guage which does not mix the dynamic elements, the uncertainty ele- 
ments, and the indeterminateness, since all these are different things that 
ought to have separate names. 

There is no disagreement of substance between Clark and myself on 
the dynamics of oligopolistic competition. I find his emphasis on the 
time interval between move and countermove as a key element in the 
theory of oligopolistic competition most appropriate and significant. 

The tangency theorem comes in for a good deal of criticism, and in at 
least one respect Clark's objection coincides with Harrod's, reproduced 
by Weintraub: businessmen are not so shortsighted as to overlook poten- 
tial competition; hence they may ward it off and secure larger sales vol- 
umes by charging lower prices. I believe that the proponents of the tan- 


gency theorem had only the case of a polypolistic seller in mind, who 
would regard himself as too small to arouse any responses on the part 
of competitors, old or new. The theorem was not meant to apply to 
the case of oligopolistic competition under the pressure of new entry, ac- 
tual or potential. 

Clark's remark on product differentiation as a competitive device con- 
firms my own conclusions that quality competition can be quite vigorous 
and that product differentiation may weaken but will often invigorate 
competition. I also showed that product standardization may be an aid to 
price competition, but is often an aid to price maintenance and other 
forms of collusion (ibid., pages 163-68). But may I take exception to the 
way in which Clark put his comment? He said that "theory appears to 
regard product differentiation as always a shelter from the rigors of com- 
petition." My theory, for one, does not. Clark obviously meant "some 
theorists." After all, no theorist is appointed or ordained to speak in 
the name of "theory." 

Gardner Ackley:! I must begin by saying that there is little in Pro- 
fessor Clark's paper with which I find myself able to disagree. As is al- 
ways the case with whatever subject Professor Clark touches, the result 
is illumination and stimulation. 

There were a few minor points out of the number he makes with 
which I might have picked an argument; but even these turned out, when 
I attempted to develop them, to become largely issues of semantics. I 
have decided, therefore, to use my few minutes for what might be con- 
sidered a methodological matter. It relates to Professor Clark's funda- 
mental thesis that our theories of competition, to be realistic and to be 
useful in policy, must be made dynamic. 

I expect that there can be no disagreement — indeed it is a common- 
place — that our economic theories need to be dynamic. The trouble is 
that the word "dynamic" does not always have a very clear or specific 
meaning. As someone has remarked, dynamic is merely an adjective 
which distinguishes one's own theories from everyone else's. 

I wonder, therefore, if it would not be useful to specify rather clearly 
what we may mean by a dynamic theory of competition before we pro- 
ceed to construct or to criticize one. I may say in advance that it is my 
feeling that Professor Clark has not given a very clear or consistent 
meaning to dynamic theory and that this may be responsible for my 
impression that his paper represents mainly a series of somewhat discon- 
nected observations. I find many of these observations individually very 
penetrating and stimulating; but nevertheless the thread which ties them 
together is sometimes hard to find. 

t University of Michigan. 


Just what is the distinction that Professor Clark means to imply in his 
contrast of "static models and dynamic aspects"? In the very first sen- 
tence of his paper reference is made to "the gap between theories of 
cost-price equilibrium and theories of production, growth, and develop- 
ment." Only a few sentences later we have it that dynamic theories "in- 
clude the tendencies toward equilibrium. But because these tendencies 
never reach their static limits, dynamic theory cannot use any features 
which are needed only to enable a model to attain this impossibly precise 
completeness. . . ." Still later we read that "a theory of competition as 
dynamic process must be not a model but a framework within which 
many models may find their places, including equilibrium models as lim- 
iting hypothetical cases." This is about all that I find which refers ex- 
plicitly to this contrast of static versus dynamic theories. There is, how- 
ever, also an implied definition of dynamic in the substantive matters and 
problems which Professor Clark chooses to discuss. 

Leaving aside for the moment Clark's particular meaning, what mean- 
ings might we give in general to this contrast of static with dynamic? 

I assume that we agree quite generally that static refers to an analysis 
of states or conditions of equilibrium. Given the values of certain outside 
or exogenous factors, there is (or may be) one pattern of the internal 
or endogenous variables from which all tendencies to further change 
would be absent. This pattern represents an equilibrium situation. 

Now, because equilibrium is a situation from which change is, by defi- 
nition, absent, equilibrium theory has frequently fallen into the trap of 
eliminating from consideration those things which are the inevitable 
concomitant of change — most of all, uncertainty. If equilibrium is a pre- 
diction — a state which will someday be reached — then this is legitimate. 
But if, as I assume it usually is, it is only an analytical device, then there 
is no excuse for eliminating from the description of behavior the effects 
of change and uncertainty. To put these back into static theory, then, 
is a first step toward making theory dynamic. 

Some of Clark's dynamics consist precisely of this. It seems to me 
that much of what he has to say about the content of long-run and 
short-run cost curves and about Chamberlin's tangency theorem is 
merely an improvement of the static analysis by the recognition of the 
effects of change and of the uncertainty which change generates. 

To introduce the effect of change and uncertainty into our demand 
and supply curves is an important step. But it is still not part of what I 
would consider a true dynamics: an analysis in which the concern is with 
processes of change. Nor does Professor Clark stop with this first step. 

As we have come increasingly to realize, in order even to define the 
existence of a stable equilibrium, we need at least some rudimentary dy- 
namics. These have always been implied in all of our equilibrium models, 
but they have often been inadequately explored. The question is: How 
do sellers (buyers, factor suppliers, et cetera) behave when equilibrium 


does not exist? What precisely is the nature and the time sequence of 
their behavior? How do they respond, and when? Only by investigating 
these questions can we be sure that an equilibrium, or at least a stable 
equilibrium, exists. And the properties of the equilibrium depend on the 
nature of the disequilibrium behavior. 

A great deal of the substance of what is now called "dynamic analysis" 
is merely an explicit exploration of the way in which equilibrium is ap- 
proached. A comparative statics analysis concentrates on previous and the 
new equilibrium positions and ignores the process of change from one 
to the other. A dynamic analysis shows the time sequence and pattern 
of movement between the two equilibrium positions. 

This type of dynamics may also recognize that full equilibrium is 
rarely or never achieved. Changes in the exogenous variables may be so 
rapid and so extensive that the system never reaches one equilibrium be- 
fore that equilibrium is replaced by another. As a well-known text once 
put it, the dog continues to chase the rabbit (equilibrium), but the posi- 
tion of the rabbit continually changes. To understand the behavior of the 
dog we need to know where the rabbit (equilibrium) is at any time, even 
though we know that the dog will never catch him. 

What would be the content of the analysis of competition which cor- 
responds to this rather narrow concept of dynamics? It would deal, I 
think, with several of the questions that Professor Clark discusses. When 
he talks of the process of profit erosion, and its duration, he is, it seems 
to me, discussing the dynamics of the approach to equilibrium. Tech- 
nological change or other disturbances to equilibrium occur. What is the 
sequence of events by which the market adjusts — through innovation, 
imitation, and competitive response — to this initial change? 

This, too, is important and useful. And it requires explicit recognition, 
as Clark urges is necessary, of the time lags in the process of market ad- 

But there is another and broader content to economic dynamics — the 
content which Clark suggests when he contrasts theories of equilibrium 
with theories of growth and development and which he deals with sev- 
eral times in the substance of his paper. The narrower dynamics retains 
an essential tie to statics; it merely investigates more explicitly the proc- 
esses involved in "tendencies toward equilibrium." But the broader type 
breaks more fundamentally with the equilibrium concept. In the broader 
dynamic analysis, movements of variables represent not merely the work- 
ing out of tendencies toward equilibrium. Rather, processes of change 
are seen at least in part to be irreversible, self-generative, and self-deter- 
mining. Change is a product of previous change, as well as of other cur- 
rent variables. Partly, this is a matter of bringing into the analysis, as 
endogenous variables, factors which the equilibrium analysis took as ex- 
ogenous. Thus we would attempt to explain, at least in part, such things 
as the state of technology, the number of sellers, the evolution of buyers' 


tastes, the nature of the market institutions (such as the form of price 
quotation and methods of sale), the attitudes of sellers toward their ri- 
vals' moves, their long-run versus short-run horizons, et cetera. These 
would be things to be explained rather than to be assumed. 

But the broader dynamics to which I refer does more than merely to 
enlarge the list of variables. It also, as I have indicated, considers the fac- 
tors responsible, not only for the levels, but for the rates of change of 
these variables. When Professor Clark sets up his criteria for appraisal of 
market performance, he is clearly thinking very specifically in these 
terms. He is concerned with whether a particular market situation is pro- 
ductive of a more or less rapid rate of introduction of new processes and 
new products, with the nature and speed of product evolution, and with 
the speed of the creation as well as the erosion of profit opportunities. 
And he has some very interesting things to say about these. 

All of these contributions toward a more dynamic theory are impor- 
tant and useful. But if we are trying to provide a general framework for 
a dynamic theory of competition, we probably need to recognize more 
explicitly than I think Professor Clark has done that "dynamic" can have 
several meanings. And I believe that we must develop somewhat sepa- 
rately the appropriate theoretical tools as well as the substantive content 
of these several levels of analysis. As is the case with other branches of 
economic theory, the dynamic analysis of competition and markets is still 
in the stage of foundation laying. 

Professor Clark has given us some sketches of a few elevations of the 
structure. But it is still a little early to tell what the building will look 
like when it is completed. 


Price Discrimination and the 
Multiple-Product Firm* 


The problem of the multiple-product firm has lain in virtual neglect 
on the threshold of the theory of monopolistic (or imperfect) competi- 
tion since the pioneering efforts of Chamberlin and Joan Robinson some 
seventeen or eighteen years ago. 1 Closely related to this problem is the 
problem of price discrimination or the price line. The significance of 
both is apparent since it is probably impossible to find in the whole of 
our economy a single firm that sells a single product at a single price. 
This is theoretically explainable by the fact that the conventional single- 

* The Review of Economic Studies, Vol. XIX (1950-51), pp. 1-11. Reprinted, 
with alterations, by courtesy of the publisher and the author. 

The writer wishes to acknowledge certain very significant criticisms of early 
drafts of this article by Professors J. M. Clark, D. H. Wallace, R. A. Lester, J. S. 
Earley, and E. W. Williams, Jr. In a broader and more indirect sense the writer has 
drawn heavily upon Professor Clark's Studies in the Economics of Overhead Costs 
(University of Chicago Press, 1923) and Professor Wallace's early but still important 
article, "Joint and Overhead Costs and Railway Rate Policy," Quarterly Journal of 
Economics, Vol. XLVIII (August, 1934), p. 583. This article, a culmination of a 
long line of articles in the Journal, just skirts the problem suggested here and 
presumably influenced and was influenced by Professor Chamberlin. The writer has 
also had the benefit of reading an early draft of Professor Earley's paper before the 
Mid-West Economic Association meeting on "The Recent Controversy on Marginal 
Analysis." Professor Fritz Machlup, who has been working on a somewhat similar 
attack on the problem, has also contributed generously by criticisms and suggestions. 

t University of Maryland. 

1 Some attacks on the problem are those of Sidney Weintraub, Price Theory 
(Pitman, 1949), pp. 289-336; G. J. Stigler, The Theory of Price (Macmillan, 1946), 
pp. 305-20; J. C. Weldon, "The Multiproduct Firm," Canadian Journal of Economic 
and Political Science, Vol. XIV (May, 1948), p. 176; and R. H. Coase, "Monopoly 
Pricing with Interrelated Costs and Demands," Economica, NS Vol. XIII (November, 
1946), p. 278. Professor R. A. Gordon in "Short Period Price Determination in 
Theory and Practice," American Economic Review, Vol. XXXVIII (June, 1948), 
pp. 273-75, likens the present concentration on the two extreme cases of the single- 
product firm and the pure case of joint costs to the previous failure to consider the 
real world existing between extremes of perfect competition and pure monopoly. 
He mentions certain other contributions on the multiple-product firm by M. W. 
Reder, W. J. Eiteman, M. R. Colberg, J. R. Hicks, and others. Kaldor has also con- 
sidered the problem. Another noteworthy contribution is that contained in the 
study of Committee on Price Determination for the Conference on Price Research, 
Cost Behavior and Price Policy (National Bureau of Economic Research, 1943), pp. 
170-88. See also Joel Dean's recent book, Managerial Economics (Prentice-Hall, 
1951), pp. 113-38, 471-548. 




product firm that is presumably in equilibrium when marginal revenue I 
is equal to marginal cost is not in equilibrium if it can serve the remain- \ 
ing portion of the demand curve at a price greater than marginal cost 
without adversely 2 disturbing its existing market, or, more commonly, if 
there is any accessible 3 market for which it can produce with its unused 
capacity at a price above marginal cost. The first situation, price discrim- 
ination, differs only slightly from the second, multiple-product produc- 
tion; together they constitute the terrain of the firm's activities. 4 

The assumption of a "product mix" does little to meet the needs of the 
situation for it lends itself to only the crudest forms of analysis and as- 
sumes away some of the most fundamental problems, including those in- 
volved in the manipulation of the firm's price and product lines. It is a 
commonplace of business practice that the production and sales man- 
agers work hand in hand to devise new products that can be produced 
with the company's idle capacity — many times at little or no profit. 
Every manager and industrial engineer is familiar with the basic tenet of 
management: What the firm has to sell is not a product, or even a line 
of products, but rather its capacity to produce. 

Any idle piece of equipment, any unused technical knowledge or or- 
ganizational resources possessed by the firm represents a challenge to the 
sales force and production manager. Any market reasonably accessible 
to the firm in which price is greater than marginal cost constitutes an 
invitation to invade. It is not necessary that the market be related to the 
firm's existing ones, although in view of management's experience it is 
desirable. Drugstores and variety stores cross-invade each other's markets 
and both invade the restaurant industry. The Du Pont Company starts the 
production of photographic film while the Eastman Kodak Company 
moves into the chemical plastics market. 

Expansion may be defensive as well as offensive. The invasion of new 
markets may have the purpose of keeping potential competitors at their 
distance. In such instances the establishment of these "outposts of com- 
petition" 5 will not be limited by the profits to be obtained. 

2 There will be some effect. See below. 

3 The word "accessible" is redundant, but it is used in recognition of the limits 
to product-line expansion which will be touched on later. 

4 Multiple-product production is universal and may be carried to extreme lengths. 
The General Electric Company is said to produce 2,000 products, The Armstrong 
Cork Company 350, and the B.F. Goodrich Company 32,000. Thorp and Crowder 
noted that at least half the products made by forty-seven out of fifty of the largest 
companies accounted each for less than half of 1 per cent of those firm's sales. 
The Structure of Industry (T.N.E.C. Monograph No. 27, 1941), p. 602. For a most 
interesting collection of cases with their ramifications see G. E. Hale, "Diversifi- 
cation: Impact of Monopoly Policy Upon Multi-Product Firms," University of 
Pennsylvania Law Review, Vol. XCVIII (February, 1950), pp. 320-56. Also C. N. 
Davisson, "Revamping the Product Line," Michigan Business Review, November, 
1949, pp. 17-20. 

5 The writer is indebted to Professor Clark for this apt phrase. Clark presumably 
used it in somewhat the same way. 


This means in theory (and apart from certain technical and institu- 
tional limitations) that production tends to be carried to the point where 
the least profitable unit of output will be produced and sold at marginal 
cost, or expressed in another way, to the point where marginal cost is 
approximately equal to price in the least profitable market. 

The rudiments of an approach to the problem of the multiple-product 
firm are already at hand in the theory of price discrimination. The strict, 
b ut useless, concept of price disrnm ination as the sale of a single prod- 
uct at seve ral prices obscures the fact that the p urposes and practices of 
price discrimination are essentially the same as those o f multiple-product 
production. What appears in the f ormer as discriminatory pr icing appears 
in tKeTatter as_acce pting different p ercentages of profit. The distinction 
between the two becomes more irrelevant when it is remembered that 
intrafirm product differentiation is often the means by which price dis- 
crimination is made possible, and is, in fact, one of the fundamental ob- 
jectives of management. The elimination or addition of a trade mark, 
o r of a few acc essories, is the means by wh ich product^ differences are 
created to the end that stron g and wea k markets can be exploited at dif- 
fe ring margins of profit. 6 D i fferentiated products merge into m ultiple 
products; price discrimination, intrafirm product diff erentiation and 
multiple-product production are more or less universal means to the same 
end and differ only in degree. 7 

Two factors have been needless obstacles to the solution of the prob- 
lem. One is the more or less tacit assumption of the inflexibility of re- 
sources within the firm. The other is the unrealistic assumption of the 
homogeneity of the firm's output. This paper assumes the transferability 
of resources within the firm and suggests a homogeneous unit of output 
to be applied to multiple-product firms. It is to be noted that the case 
of true joint costs and invariable product proportions is thus specifically 

Any study of industrial techniques indicates that the transferability of 
resources between products and within the firm is much greater than is 
commonly assumed in economic treatises. 8 Lathes, milling machines, 

6 As an interesting case in point note the varying profit margins on the several 
cars produced by General Motors Company and the Chrysler Corporation. See 
Federal Trade Commission, Report on the Motor Vehicle Industry (1939), 
pp. 538, 603. 

7 The problem discussed here is closely related to the problem of marginal cost 
pricing which the writer touched upon in another article. E. W. Clemens, "Price 
Discrimination in Decreasing Cost Industries," American Economic Review, Vol. 
XXXI (December, 1941), p. 794. The most recent and comprehensive discussion of 
the whole marginal cost controversy is that of Nancy Ruggles, "Recent Develop- 
ments in the Theory of Marginal Cost Pricing," Review of Economic Studies, 
Vol. XVII (2), No. 43, 1949-50, p. 107. 

8 This principle, as well as the others embodied in this article, are illustrated in 
any standard text on industrial management. Particularly significant are the cases in 
F. E. Folts, Introduction to Industrial Management (McGraw-Hill, 1949). These 
cases can be assumed to represent a fair cross-section of industry. 


drill presses, boring machines, etc., can be shifted at will between a large 
variety of products. At the most only a new set of jigs, dies, patterns, 
etc., are necessary. Nor is it necessary that all equipment be generally 
adaptable; it is only necessary that the firm have a reasonable proportion 
of general-purpose equipment (as distinguished from special-purpose 

The same principle holds true in the processing industries. The pro- 
portion of oil refinery products can be varied. It would be completely 
unrealistic to think of the equipment of a chemical company to be spe- 
cialized by products. Most nonferrous metal companies produce a vari- 
ety of metals, coal is produced in varying grades for different markets. 
A steel company produces a thousand types of steel, each to be sold in 
different markets, with the same facilities. The transferability of re- 
sources between markets is, of course, commonplace in the railroad and 
utility industries. 

The problem of homogeneous units of output can be solved within the 
conventional framework of the theory of monopolistic competition. 
Units of output are defined as blocks of output, without distinction as 
to products, which have equal direct costs under standard conditions. 
This is more than a mere system of weighting; it corresponds closely 
with data obtained from motion and time studies and made a part of 
permanent shop and production records which in turn are used in man- 
agement decisions. These direct costs would, of course, differ from both 
anticipated and realized marginal costs which would vary over the range 
of the company's output. 9 With units of output so defined it would be 
desirable generally to assume rising marginal costs as less efficient, stand- 
bv equipment and labor are brought into use. 


The assumptions of partial (although not necessarilvc omplete) tr ans- 
ferability ot resources within the firm be tweerTTprod ucts and a homoge- 
neous unit of output basec Ton standard~cfire ct costs are both realistic and 
theoretically prac ticable. The problem of the multiple-product firm can 
then be treated simply^as a problem i n price discrir mnation, ancT loan 
Robinsons w ell-known analysis of what Pigou ^ajls_ d^riminationj ofthe 
thir d degree will lend itself with cert ain reservations to the considera- 
tion of the problem. 10 The analysis, however, is subject to^eTtamlmpor^ 
t ant limit ations i f the~attempt is made 10 apply il lu llie'^a^ le^pxoxiuct 
firm , a use presum a bly not contemp lat ed by Mrs. Robinson^ 

9 Direct costs under standard conditions are analogous to what operating man- 
agement commonly calls "standard costs" which reflect what an article "should 
cost under normal operating conditions." An important function of operating man- 
agement is the analysis of deviations from these standard costs. See C. C. Balderston, 
V. S. Karabasz, and R. P. Brecht, Management of an Enterprise (Prentice-Hall, 
1942), pp. 380-84, for example. 

10 Joan Robinson, Economics of Imperfect Competition (Macmillan, 1936), chaps. 
15 and 16. 


In the first place, the Robinsonian method does not visualize the ex- 
tension of production to the point where marginal cost is equal to de- 
mand. The market and its several components are clearly defined. It only 
remains for the monopolist to maximize his profit in each market. Nei- 
ther the dynamic nature of the firm's activities nor their scope is revealed. 
If it is assumed that the firm could reach a large number of markets in 
which price is greater than marginal cost, the firm could not be in equi- 
librium. Another point of necessary refinement is that the Robinsonian 
analysis does not include the situation where the firm can distinguish be- 
tween units of output sold to customers in the same market. Thus a firm 
might apply a sliding scale of prices (or quantity discounts) to each mar- 
ket. Customers might be classified not only on the basis of the elasticity 
of demand of the group, but simply on the basis of the maximum price 
each customer is willing to pay for a certain number of units. 11 The in- 
troduction of such an assumption might make the Robinsonian technique 
completely unworkable because of the sheer complexity of the resulting 

^ Another limitation of the Robinsonian analysis for the present case 
has to do with her use of the conventionalized marginal cost curve by 
which marginal cost is applicable to any increase in output regardless of 
form. In many cases of empirical research it is desirable, if not absolutely 
necessary, to identify marginal cost at any particular level of output 
with specific increases rather than any increase in output. This is not 
possible by Joan Robinson's technique. 


The procedure suggested here is believed to be somewhat more ap- 
plicable to empirical research and is also consistent with certain more 
dynamic aspects of the firm's activities which are not considered by Mrs. 
Robinson. Co nditions ranging f rom strong mo nopoly to virtually perfect 
c ompetition are assumed in the various markets t hat_can be reached by 
th e company . Practically, competition is assumed to take the form o f in- 
vading ne w markets by a process of pri ce discrimination. Price d iscrimi- 
nation thus emerges parad oxically as the most common means of compe- 

It is as sumed tha t the firm's resources are mobile and that the firm 
can produce a wide variety of products. The case of joint costs and fixed 
product proportions is specifically excluded, althou gh th e procedure 
c ould be modifi ed~t^in^lua r e such cases. It is also assumed that deman ds 
are not related. This is consistent with the objectives of the firm, since 
the higher the degree of market independence, the greater become the 

11 The typical block rate schedule of the gas and electric utilities is an example. 
See K. E. Boulding, Economic Analysis (Harper, 1941), pp. 540-49. For another 
attack on the problem see William Vickrey, "Some Objections to Marginal Cost 
Pricing," journal of Political Economy, Vol. LVI (June, 1948), p. 218. 



advantages of (and the possibility of) breaking up the price structure of 
the total market. It is also true that the further we move from price dis- 
crimination towards the concept of the multiple-product firm, the more 
reasonable the assumption becomes. Furthermore, the larger the number 
of firms and the more diversified the markets of each, the smaller will 
become the consequences of the intermarket relationships of each. 

The firm might-weiLJie_ assmueaLjo_sta£tL-as_-^--conventional single- 
product firm operating under ^conditions of 60 or 70 per cent of capacity 
and at a point where marginal cost equals marginal revenue. Possessing 
excess capacity in the form of equipment, personnel, and organization, it 

can increase production without an undue increa se^ _in_ marginal cost. 
Rather than reduce the price in an existing market it will seek new out- 
lets in wju^cji_de mand exceeds marginal costTT t will be assumed^or~pur- 
poses of analysis that new markets are invaded in order of their profit- 
ability. The firm wi ll be in equilibrium insofar as its own management is 
concerned only whenthere are no more accessi ble markets in which de - 
mand price is greater thalTmarginar cost. The situation is illustrated in 
Figure^ 1. The figure assumes five markets, although the markets might 
well be innumerable. Profits will be maximized when production is dis- 
tribute dbetw£eri_riie_^yemar^ marginal 
revenue equal in all markets and equal to marginal cost. The E MR line 
is a "I me^bTequal m arg inal revenue. 

In contrast to Joan Robinson's method, each market has its own zero j 
output axis which is the source of that market's marginal revenue curve. 
Demand is assumed to be a chain of demand curves, the EMR line de- 
termining output in each market. If the analyses were started with the 
assumption of a certain and limited number of accessible markets, the 


| EMR line would have to be determined by Joan Robinson's method — 
that of adding horizontally the marginal revenue curves of all the mar- 
kets to obtain the aggregate marginal revenue curve. The intersection of 
this curve and the marginal cost curve would determine aggregate output 
and the position of the EMR line. This AMR curve is not shown, but a 
segment of Joan Robinson's aggregate demand curve, AD, is shown at 

j operating output. 

Five product prices are established, the lowest being just in excess of 
marginal cost. Although it is assumed that the markets are aligned from 
left to right in order of their chronological entry, the corresponding as- 
sumption that they were entered according to their profitability results 
in an alignment according to three standards: (1) inversely according to 
the elasticity of demand in each market, (2) according to the margin of 
profit measured as a percentage of either standard direct costs or price, 
and (3) according to price per standard unit of output. 

Putting it somewhat differently, xht EMR lin e is set at a level deter- 
mined by the intersection of the firm's marginal "cost curve ancT the MR 
curve for the last market that can be served profitably. In theory this 
market is one with the most elastic demand, and the price established 
for it would be barely in excess of marginal revenue and marginal cost. 
This brrngs^bout practical equivalen ce of marginal cost, marginal rev e- 
nue, and demand in the marginal market. 

If this last market is one of perfect elasticity, equivalence is perfect. It 
also follows that if any single market demand served by the firm is one 
of infinite elasticity, it becomes the marginal market. Any marginal mar- 
ket of less than infinite elasticity leaves open the possibility, although 
not the absolute necessity, of some unserved market in which price is 
greater than marginal cost. 

We can inte rrupt the train o fth e argu ment to make the same basic 
point by use oitjoan Ro binsonV technique^ Figure 2 is adapted from_h ex- 
Figure 61. 12 She has assumed the firm to be produc ing an output OM de- 
termrneanSyntKe int ersection of the marginal cosTcurve with the aggre- 
gateTmarginal revenue curve AMR. PnVpg in two rnark ets are Pi and P 2 . 
T heTJrrfi, "however, couTcT not be in equilibrium if there are any other 
accessible markets in which price is g reater than CM (and presumabl y 

•" less than P 2 ) . If we assume a series of such markets with their elasticities 
of demand ranging upward by degrees to infinity the marginal market 
or least desirable market would be either one of infinite elasticity or an 
extremely small market with price just above marginal cost. Under any 
other circumstances the divergence of price in the marginal market from 
marginal cost would make further diversification profitable. Of several 
possible markets of perfect elasticity, the limiting one would be the one 
offering the highest price for the standard output unit. The others would 

12 Op. cit., p. 182. 



be precluded from consideration since there would be no need to shift 
production to a lower price market if the firm could sell all it desires at 
a higher price. 

A limiting market of perfect elasticity, D 3 , and its marginal curve, 
MR 3 , is shown in Figure 2. For purposes of simplicity it is assumed that 
this market is the only remaining accessible market in which price is i 
less than F 3 and greater than CM, although the preceding argument as-/ 
sumes a series of such markets. The horizontal addition of the single or- 


dinate to the existing D and AIR curves breaks the AMR and AD curves 
at their respective intersections with the horizontal D S -MR 3 line of which 
AMR 3 and AD 3 are continuations. Production is stabilized at an output 
0-M 3 . Additional output will be sold at a price P 3 in the marginal market. 
Prices P 1 and P 2 will be raised and the output in these markets will be 
further restricted. 

If it is to be assumed that there is no such market of perfect elasticity 
of demand m wh ich_pn ce exceeds marginal cost, the limiting market 
would be the one with the greatest elasticity. I f trie assumption of a w ide 
variety of markets is to be maintained, the limiting one in this case would 

have to be relatively small and would afford a price only a little i n ex- 
c ess ot marginal cost. Ex tension to this market would probably involve 
extreme diversification. The practical limits to this process will be dis- 
cussed later. It is also interesting to note that the limiting market might 
be an unserved and highly elastic segment of one of the existing seg- 



ments if it could be broken away from the existing high price-inelastic 

To return to Figure 1, the firm would not be in equilibrium if there 
were any other market open to i t for which any part of the dem and 
curve would lie above the EMR line. The addition of any such market 
would result in "squeezing" production in the firm's existing markets. 
Output in each of them would be slightly restricted and a new EMR 
line would be established at a level slightly above the old one. The AR 
curve would move to a new and higher position. If marginal costs had 
begun to rise rapidly (if the firm had been operating at "capacity") the 
sudden appearance of a new and profitable market might have the jeffect 
of raising the EMR tine so far as to "submerge" the demand curve for 
one or more products. These products would then be crowded out of th e 
company's line. The situation was of common occurrence in the sellers' 
market following World War II when many low-profit items disappeared 
from the market. It is to be noted that the extension of production to 
the point where marginal cost is equal to demand increases total output 
but encourages the restriction of output in individual markets. 

The marginal cost curve is assumed to be continuous although the way 
is open to kink it at output breaks. This would facilitate the application 
of theory to many practical problems of analysis. If the marginal cost 
curve is assumed to be unaffected by the nature of the firm's products, 
both the average and marginal cost curves would be similar in nature 
and comparable to conventional curves drawn for the single-product 
firm. On the other hand, the values of the average revenue curve would 
be applicable only to a designated order and proportioning of the various 
products, except at the point of operating output. 13 

The activities of most firms would indicate an assumption of an almost 
unlimited series of demand curves. In many instances the need of anal- 
ysis could be met by classifying customers into a few groups according 
to prices per standard unit of output and the inelasticity of demand of 
each group. Creation of iso-elastic demand groups would involve the 
surrender of product identities; but in cases where the sacrifice could be 
made, analysis would be simplified for many types of problems. 14 

The market alignment procedure suggested here reflects the dynamic 
expansion process of the firm which constantly seeks greater profits bv 

13 The shape and level of the marginal cost curve would also be influenced by 
the composition of the standard units of output and the standard direct costs assigned 
to each of the products. Standard costing techniques would furnish the basis for 
the distribution of costs. It is also to be noted that nothing would preclude es- 
tablishing individual MC curves, each one beginning at the product's zero axis and 
intersecting the product's MR curve on the EMR line. The MC curve would be 
the horizontal sum of these product MC curves. 

14 Professors Walter Rautenstrauch and Raymond Villcrs in their recent Budgetary 
Control (Funk & Wagnalls-Modcrn Industry, 1950), chap. 9, have outlined this ap- 
proach as a management device. Products are classified according to the percentage 
of profit on sales and without regard to their nature. 


succe ssively invading less and les s profitable markets. Market alignment 
according to inelasticity of demand has several technical advantages. 
Attention is focused on the spatial nature of the total market and upon 
the marginal market. Analysis is simplified and comparisons may be made 
easily with the conventional price -cost relationships of t he single-product 
firm. Finally, specific segments of the marginal cost curve are tied to par- 
ticular blocks of output. This is consistent with accounting procedure 
and with the thinking of management which is generally in terms of 
blocks of output, such as job lots or products, rather than in terms of 
units. It can be doubted whether in practice marginal cost has any sig- 
nificance at all except in connection with specific job lots of products 
under the conditions which are peculiar to the specific line already pro- 
duced. 15 

The virtual equivalence of marginal cost, marginal revenue and de- 
mand in thelnarginal market merely recognizes the fact that practically 
all firms produce some products at little or no profit which are just on 
the margin of being dropped. In other instancesno profit items are pro- 
duced in o rder to keep t he organization intact. The practice has long 
been taken for granted in the railway and utility industries. Selling at 
marginal cost by freight absorption is another example. A study of pro- 
duction, marketing and cost accounting techniques offers many exam- 
ples. Basically the procedure is one of restricting output in inelastic mar- 
kets while avoiding the costs of idle capacity. However general this 
practice might be, it seems indisputable that the divergence between 
demand and marginal cost at operating output is much less than is com- 
monly assumed for the single-product firm. 


The question can well be raised as to whether there are limits to the 
process of market invasion. In view of the countless number of products 
produced (or sold) by the typical firm and the steady drift towards large- 
scale enterprise, the answer in both theory and practice might well be no. 
But obviously there are limits. There are technical limits to the number 
of products to which plant capacity can be adapted and physical limits 
to the markets which can be reached. The most common limits are 
probably those produced by the economies of specialization. In recogni- 
tion of these economies management characteristically sets up certain 
minima for output blocks. 16 Thus it can be said that regardless of how 
many products a plant is physically capable of producing production 
economies and diseconomies will establish a much more restricted limit. 

15 Thus the marginal cost of producing product No. 25 might depend upon the 
utilization of waste parts from products Nos. 5, 11 and 14, rather than upon general 
capacity conditions. Here, of course, there is an element of jointness. 

16 For a discussion of management procedure in establishing lot sizes see L. P. 
Alford, Cost and Production Handbook (Ronald, 1934), pp. 236-42. 


To make the same point in a different manner, firms may often operate 
at a point where any substantial increase in output will bring sharply ris- 
ing marginal costs. Rising costs will be even more pronounced if expan- 
sion is to be achieved by the invasion of a new market. The marginal 
costs of invasion are likely to be considerably higher than the marginal 
cost of expansion by the firms already in the market. Among the more 
easily identifiable of these costs are those of transportation to distant 
markets, of missionary sales work, and of tooling and machine set-ups. 17 

To the extent that firms base their price and production policies on 
average rather than marginal costs, expansion will be restricted. The 
writer shares the dissatisfaction of Professor Gordon 18 and others with 
the marginal cost function, but it is more than conceivable that its vul- 
nerability centers more in the imputed precision than in its substance. A 
few observations are relevant without being conclusive. Generally, firms 
price many products closer to their concept of marginal cost than would 
be indicated by the marginal cost concepts that emerge from many eco- 
nomic treatises. Marginal cost is something more than its ascertainable 
and measurable elements. Risk and the additional cost of management, 
both of which are substantial, are marginal costs. Conceivably marginal 
cost must include a certain minimum amount which the businessman 
considers necessary profit. When he says that the profit on a certain ad- 
ditional piece of business is "not worth his time and trouble," he is giv- 
ing expression to a very real concept of marginal cost. The writer also 
agrees with Professor Machlup's position that too rigid a definition of 
marginal cost would trap its users into unrealistic and untenantable posi- 
tions. 19 

Cost accounting techniques indicating an average cost approach to 
price and product line policies have perhaps needlessly worried some de- 
fenders of marginal economics. Typically, the determination of average 
costs (or standard costs plus a margin for overhead and profit) is a func- 
tion of the cost accountant in the lower echelons of management. Cost 
analyses, however, represent only the basic data from which price and 

17 Tooling, machine set-up, and similar costs are direct and marginal costs before 
a market has been invaded. After invasion they are, of course, fixed costs. These 
new fixed costs must either be lumped with the firm's general fixed costs which 
would lower the marginal cost segment for the particular product after invasion, 
or they might continue to be considered as part of the product's marginal cost. 
The question is considered in Professor Austin Robinson's review of Manufacturing 
Business, by P. W. S. Andrews (Macmillan, 1950) in the Economic Journal, Vol. 
LX (December, 1950), pp. 771, 778. Both the book and the review are relevant to 
the problem considered here. 

18 See R. A. Gordon, "Short Period Price Determination in Theory and Practice," 
American Economic Review, Vol. XXXVIII (June, 1948), p. 265; with other cita- 
tions to basic contributions by R. A. Lester, Fritz Machlup, G. J. Stigler, and 
H. M. Oliver, Jr.; see also Lloyd G. Reynolds, "Toward a Short-Run Theory of 
Wages," American Economic Review, Vol. XXXVIII (June, 1948), p. 289. 

19 See Fritz Machlup, "Marginal Analysis and Empirical Research," American 
Economic Review, Vol. XXXVI (September, 1946), p. 519. 


production strategy is plotted in light of other factors by top-flight man- 
agement. In different terms, average costs are significant to those in the 
management hierarchy who follow policy, but not necessarily to those 
who make it. To top management some circumstances might dictate 
pricing or the addition of a product at only a little above what the cost 
accountant's statement indicate to be marginal costs. Other circum- 
stances might lead management to reject suggested additions to the prod- 
uct line that cover average costs several times over. To some extent the 
solution of the problem and the controversy involved turn on the pe- 
riod assumed for analysis. The longer the period for which strategy 
must be plotted, the greater becomes the percentage of total costs which 
must be characterized as marginal. 20 

One of the most important obstacles to expansion into new markets is 
the lack of "know-how" 21 on the part of both management and workers. 
The fear of prosecution under the Sherman Act has been significant in 
many cases. On the other hand, tacit market-sharing agreements may 
limit invasion. Such agreements are found even among small-town mer- 
chants. For these and other reasons we must assume that there are bounds 
to the number of accessible markets. But whatever the bounds, firms 
will push production in varying_degr ees towards the limit where mar- 
gini rcostTs^equal to demand in the marginal market. Profit jria rgins in 
the_yary[ng markets will differ due to localized monopolies, lags in c om- 
petition and other factors. T o assume conditions of competitive equi lib- 
rium whe re all profit margins are equal would be completely unrealistic. 
No rmal "profits, necessary to a firm's long-run exist^nceTare ob tained 
only insofar as average revenues under multiple-product production are 
equal t o averag e cos ts. This co ndition can only be att ained by the con- 
tinuous process of invasion and cross-invasion of markets, by the shuf- 
flfnglind reshuffling of prices and markets, which are so cha r acteristic of 
economic activity. 22 


Ajur ther p oint, generally neglected in the theory of price discrimina- 
ti on, will round o ut the discussion. In some instances it can be assumed 
th at the fi rm has the power to transfer^articular~customers or demand 
units between price groups. In other words, the firm might not only 

20 The writer has greatly benefited from correspondence with Professor J. S. 
Earley in the development of the preceding material. 

21 An increasingly common method of overcoming this obstacle to the invasion 
of new markets is the pooling of technical and other resources in the hands of a 
firm sponsored jointly by parent firms in two or more industries. General Electric 
Company, Pittsburgh Plate Glass Company, Corning Glass Works and many of the 
large rubber and chemical and other companies have participated in the formation 
of such companies. 

22 C. N. Davisson, op. cit., p. 1, n. 5, discusses some of the advantages and limita- 
tions of diversification under existing business conditions. 



have the power of differentiating between customer classes^ but it might 
have the further power of determining the constituency of each class. 
Customers nominally in Group 2, but willing to pay more than P 2 , might 
be shifted to Group 1 and charged a higher price. Unsatisfied customers 
nominally in Group 2 might be s hifte d to a lower price group. To the 
extent that this can be done th^ chainj a f demand cu rves tends to merge 
into a single continuous demancTcurve. 

In such an instance the monopolist might select a serie s of pri ces along 
the demand curve, each price being applicable to all customers willing to 
pay it and unwilling to pay the next higher price. In short, the demand 
curve would be "segmentized." The more prices that could be estab- 
lished, the greater would be the monopolist's profits. In the extreme and 
' J * limiting case the monopolist would ob tain the full demand price f or each 
unit of output by establishing an al most infinite series of prices. Con- 
sumers' surplus would be entirely eliminated and the demand curve 
would become the marginal revenue curve. 23 

First thought might proclaim such assumptions to be inconceivable. 
However, our economy furnishes numerous cases where the procedure 
could be applied with a high degree of accuracy. If t he firm is able to 
break up the market into submarkets and apply sliding scales of quantity 
discounts to each customer group, the situation will be approached. This 
depicts the pricing pattern of gas and electric utilities where firms dis- 
criminate not only between classes but between units of output taken by 
individual customers within each class by block-rate schedules. 24 The 
method is also applicable to those firms that make it a policy to succes- 
sively exhaust demand at lower and lower price levels. 25 It would be 
applicable to a monopolistic custom-order firm that set prices on a 
negotiated basis. One of the more interesting applications would be to the 
increasing number of patent-owning firms who license the use of pat- 
ented products, the license payments being graduated according to the 
revenue, or profits, obtained from the use. In general the method would 

23 A completely segmentized demand curve is the same as Pigou's discriminating 
monopoly of the first degree. A partially divided curve represents his discriminating 
monopoly of the second degree. Discriminating monopoly of the third degree is 
the basis of Joan Robinson's treatment. Commonly the latter case is the only one 

24 Professor M. G. Glaeser, following Taussig, Pigou, Edgeworth and others, used 
the technique in analyzing the theory of utility rate making. Outlines of Public 
Utility Economics (Macmillan, 1927), pp. 618-39. In consideration of the extent of 
the power of classification possessed by utility companies and the multiplicity of 
their rate schedules this simplified approach to the problem is much more accurate 
for most purposes than Joan Robinson's more refined method which does not con- 
sider intraclass discrimination. The method is also discussed and applied to a limited 
extent b)C^roTesso"r r Boulding. Economic A nalysis (Harper, 1941), rjrj ^iJjQ^^^ 

25 As, for example, in department store selling where shelves are cleared by suc- 
cessive price reductions, or in the publishing field where cheaper and cheaper edi- 
tions are brought out, or where a new product is introduced at a high price which 
is steadily reduced. 



facilitate analysis of a multitude of firms that practice discrimination to a 
considerable degree and whose activities cannot be analyzed conven- 
iently by either the single-firm or the Robinsonian technique. The tech- 
nique certainly must be used cautiously, but the areas of potential ap- 
plication seem to indicate that it could be included justifiably in the 
economists' traditional "kit of tools." 

Figure 3 has certain theoretical implications. It will be noticed that 
the average revenue curve is not the same as the demand curve but lie s 
above it. Hence, it is possi ble under these circumstances for a firm to 

! ^^5l 

i ' n** — n 


o 2 o 3 



Figure 3 

make normal profits even if the demand curve lies continuously below 
the average cost curve. 26 This point has sometimes been stressed in the 
theory of railroad rate making. 

Itis also to be noticed that the divergence between the demand and 
average revenue curves is increased as the demand curve is segmentized. 
Management achieves maximum profits in this conne ction in two ways: 
(1) by invading new markets and (2) b jCspjintering^itg exjsjdng juarkets^ 
But whether the fractionalization of the firm's market is extensive or in- 
tensive, the end result is the same. 

The divergence of the demand curve from the average revenue curve 
is not confined to this special case. It would follow in the first and more 
general case that only in the rare circumstances where the elasticity of 
demand was uniform in all of the firm's markets, i.e. where the percent- 
age of profit on price or standard cost was the same, would there be any 

26 Entrepreneurs contemplating the establishment of a new firm seldom if ever 
make their decision by setting an assumed price against an assumed cost of a single 
product. If such were the case very few new firms would ever appear. Rather the 
process is one of comparing the total revenues from a series of products with the 
total costs. 


identity between the average revenue curve of the firm and its demand 
curve, however conceived. 27 


The foregoing a nalysis should suggest that pri ce discrimination and 
multiple-product production are not exceptions to "general practice, b ut 
are rather the essence of customary action. T he dis tinction between a 
producer selling a single product at different prices and one selling dif- 
erent products in varying markets at differingjercentages of profit is 
a distinction of degree only^What a firm has to sell is not a p roduct, but 
rather its capacity to produce. Insofar as f irms are motivated by the mar - 
ginal principle, there is a tendency to push production towards the point 
where marginal cost is equal to demand price for the least profitable unit 
produced. In this manner "capacity" operation is achieved. The limits to 
- the process are InsHtutional rather than theoretical. Under conditions of 
monopolistic competition this probably results in a situation where only 
normal profits are obtained. Whatever the amount of profit, it is obtained 
only by constant manipulation of the price and product line. The theory 
of price discrimination must be viewed as the heart of price-cost theory 
r ather than a peri pheral case. T he firm that doesnortrsmrrmTate Tin Its 
pricing policy, or differentiate in its product line, or invade newjmarkets, 
dies in the competitive struggle — and "busine ss management do es not 

27 It is extremely questionable whether the ordinary concept of "the demand 
curve of the firm" can be applied at all in the real world of multiple-product 
firms. In this instance it is assumed to be the same as Joan Robinson's aggregate 
demand curve (in terms of standard units) which is obtained by adding the various 
demand curves horizontally. 



The Influence of Market Structure 
on Technological Progress 


The argument of the present article will be developed by gradually 
relaxing the simple but unrealistic initial assumptions. This procedure has 
obvious disadvantages as well as advantages. The writer feels that in this 
case the advantages outweigh the disadvantages because certain familiar 
propositions pertaining to our problem imply the early assumptions of 
the article, and it seems useful to construct a bridge from these proposi- 
tions to the later conclusions of the analysis. The titles and the intro- 
ductory sentences of the various sections describe the assumptions in a 
general way but two points should be added here. In the first place, the as- 
sumption that inventions and know-how are immediately available to all 
potential users will be removed only in the final section of the article. 
Secondly, the assumption that progress (the growth of output) is asso- 
ciated with falling prices rather than with rising rates of money income 
will not be removed at all. From our point of view no importance attaches 
to the difference between these two ways of distributing the fruits of 


In this section it will be assumed that we are faced with a single un- 
expected invention, and that entry is free in each industry, in the sense 
that there exist no institutional (discriminatory) barriers. The second 
condition is compatible with real economies of scale such as might place 
obstacles in the way of entry. The potential entrant might know that 
firms smaller than the existing ones would be less efficient and that he 
therefore would merely have a chance of replacing an existing producer. 
This might prevent him from intruding. If such obstacles exist, we shall 
not say that entry is "blocked" but it is to be expected that in these cir- 
cumstances the typical results of free entry will show in a somewhat 
qualified form. Furthermore, if in consequence of product differentiation, 

* The Quarterly Journal of Economics, Vol. LXV (1951), pp. 556-77. Re- 
printed, with alterations, by courtesy of the Harvard University Press and the 

t Yale University. 




entry is free merely into a Chamberlinian group, then "free entry into an 
industry" becomes qualified by buyers' preferences. The concept "in- 
dustry" loses its precise meaning in these circumstances. Discussion of 
this problem will also be included in the present section — entry will not 
be regarded as "blocked" — but here again the assumptions of the section 
will be said to possess merely qualified validity for the problem. 

For a purely competitive industry the situation is portrayed in Fig- 
ure 1. The long-run supply curve of the industry (LS) is represented as 
horizontal. This does not influence the result of the analysis. Magnified 
cost curves of the single-plant firms are ATC, AVC and MC. The in- 
dustry demand curve is DD. The equilibrium output is OA, the equilib- 
rium price OL. The number of firms (plants) is determined by the con- 

dition that at the equilibrium output (OA) all firms operate at minimum 
average total cost. 

The short-run supply curve of the industry (SS) is the sum of the 
AVC minima up to point C on the abscissa where the output is suffi- 
cient to keep all existing firms in operation at minimum average variable 
cost. At output OC the short-run supply curve turns up, expressing the 
fact that at prices higher than OS, each firm will increase its output along 
an individual supply curve which is its marginal cost curve. Starting from 
the equilibrium output of the industry (OA), the (from our point of 
view) relevant supply curve is short run to the left and long run to the 
right. This is because we are concerned with problems of entry and our 
analysis will be sufficiently long run to warrant the assumption that an 
increase of output beyond OA will be brought about by the entry of 
new firms. However, our analysis will not be sufficiently long run to lead 
us to disregard the fact that a reduction of output remains compatible 
with an unchanging number of plants during a period of substantial 

Let us now assume that the equilibrium becomes disturbed by an in- 


vention that lowers the long-run cost curve from the LS level to LSi. 
This lies between the average total cost and the average variable cost of 
the old firms. New firms will enter with the new method along the new 
long-run function LS±. The price reduction will force the old firms to 
move back along the short-run function. The lowering of the long-run 
cost curve from LS to LSi raises the equilibrium output of the industry 
to OB, and it lowers the equilibrium price to OL x . The aggregate output 
of the old firms will be OD and entrants will fill the gap between OD 
and OB. Whether the old firms do or do not change their methods of 
production immediately depends on whether the innovation passes a test 
which will be described in Section II and will there be called the Figure 3 
test. We have assumed that in this case the test is not passed: owing to 
their fixed costs, the old firms find it more profitable to continue with 
the original technique. 1 Even if they do not introduce the new method 
immediately they will have to reduce their selling price to that cor- 
responding to the new method. This of course inflicts losses on the old 
firms. After a period that may be of substantial duration, all firms 
will become located along LSi and all will be using the new technique. 
This is because "in the long run there exist no fixed costs." As we shall 
see in Section II the "Figure 3 test" would always be passed, were it not 
for fixed costs. 

So far atomistic competition and a homogeneous product were as- 
sumed. As long as there exist no institutional (discriminatory) barriers 
to entry, the conclusions remain similar. However, some modifications 
are required to allow for the special characteristics of various market 

In monopolistic competition in the large group new firms enter with 
new "brands" that could not have been produced profitably prior to the 
cost-saving invention. If there exist differences between the popularity 
of the various brands, not all old firms need fully to adjust their prices to 
the cost saving. This is a significant difference but not a fundamental 
modification of our conclusions because the difference is a consequence 
of the fact that in this case the freedom of entry is imperfect. The in- 
dustry concept ceases to be precisely applicable to the problem because 
unqualified freedom of entry must mean freedom of entry into an "in- 
dustry" producing a homogeneous product. The previous conclusions 
will be neither more nor less valid than is the concept of "free entry" in 
cases where no firm can produce a perfect substitute of its competitors' 
product. If new firms can enter merely somewhere in the "neighbor- 

1 It will be seen later that the failure of the innovation to pass the test expresses 
itself in the fact that the lowering of ATC to tangency with L5, does not make 
this point of tangency lie below the A VC curve. The A VC curve of already existing 
firms may, in the first approximation, be conceived of as remaining unchanged. 
More precisely, it should be regarded as gradually shifting upward, owing to the 
wear and tear to which the plants are exposed. 


hood" of the old rather than in the identical industry — the neighborhood 
being defined in terms of cross-elasticities — then the degree of price ad- 
justment which is forced upon the old firms depends on how "close" to 
each of the old firms the entrants become located. This in turn is a matter 
of the distance of the various products from one another in economic 
space (as defined by cross-elasticities). It is a matter of how strong the 
monopoly element is in the monopolistic competition. The smallest cost 
reduction will induce entries if the products form a continuous array. If 
there exists a discontinuity in the neighborhood of the margin, i.e., at the 
old equilibrium, then more than a very small cost reduction is required to 
induce entry but all these modifications of our conclusions stem essen- 
tially from qualifications of the assumption of free entry into an industry. 
In these circumstances it is impossible to engage in the production of 
(practically) the same commodities as those produced by the existing 

Monopoly may of course develop even if it is possible to produce 
(practically) the same commodities as those already being produced. 
Such monopoly is compatible with our assumptions concerning freedom 
of entry, as long as the monopoly position stems exclusively from real 
economies of large scale. For such monopoly our conclusions hold with 
a modification which follows from the cost advantage enjoyed by the 
monopolistic firm. The price charged by such a monopoly cannot exceed 
the level at which atomistic competitors would find it profitable to enter 
but it can fall short of this price. 2 If it falls short of it, then the inven- 
tion must reduce the costs of the potential atomistic entrants by more 
than the difference in question to force the monopolist to adjust his price. 
In other words the conclusions which were established for pure com- 
petition follow here only if the cost reduction makes the latent competi- 
tion of atomistic outsiders effective. The latent competition of nonatom- 
istic (i.e., large) potential entrants need not affect the price because these 
should be assumed to know that their entry would reduce the price and 
that it would also reduce the firm size below optimum. They have no 
reason to enter unless they think that they can ultimately displace the 
existing firm. The availability to them of an innovation which is avail- 
able also to the existing firm does not justify this belief. 

In conclusion we may say that on the assumptions expressed by the 
title of the present section, the smallest possible improvement will induce 
entry, price adjustments, and growth of output, provided firms are atom- 
istic and the product is homogeneous. If the product is not homogeneous, 
the free-entry assumption becomes qualified and this qualifies the con- 
clusions. If, owing to economies of scale, monopoly exists in spite of 

2 In Professor Bain's terminology, the limit price is on the level where entry 
would occur. But the intersection of marginal revenue with marginal cost may 
occur at an output level for which the price is below the limit. Cf. Joe S. Bain, 
Pricing Distribution and Employment (New York, 1948), pp. 189 rT. 


free entry, price may be lower than that which would induce small firms 
to enter and this too qualifies the conclusions. 


The discussion will now turn to the conditions which must be satisfied 
before an existing firm adopts an invention as an innovation. In the pre- 
ceding section the existence of such conditions was merely hinted at be- 
cause the results were not determined by whether the old firms did or 
did not promptly adopt a new method. The results were determined by 

Figure 2 

entry. If entry is blocked the results follow directly from the behavior of 
the existing firms. 

The problem to be discussed in the present section is intimately con- 
nected with the relationship between the short- and the long-run mar- 
ginal cost functions (SM and LM) as portrayed in Figure 2. The es- 
sential property of the graph is that the short-run marginal cost is lower 
than the long run to the left of the equilibrium (OA) but higher than 
the long run to the right of it. This expresses the fact that the plant is 
such as is called for by output OA with correct anticipation of a mar- 
ginal revenue curve such as MR. Incremental cost for an increase in out- 
put is higher along SM (i.e., given the plant) than would be the case if 
the plant had not yet been built for OA output (or indeed for any out- 
put at all), and we would be comparing total cost for OA with those for 
higher outputs, assuming the correct plant for each output. On the other 
hand, decrease in total cost is smaller along SM (given the plant cor- 


responding to OA) than along LM, for the analogous reason. SM could 
be horizontal and LM downward sloping, or both cost curves could be 
downward sloping, or LM could slope downward or run horizontally 
and SM slope upward in the neighborhood of the equilibrium. But SM 
would still have to intersect LM from below, and the intersection would 
have to occur at the long-run equilibrium output. 

The decision to expand along LM rather than SM requires that at the 
new equilibrium the total area lying under LM, computed from the ori- 
gin, should be smaller (at least no greater) than the total area lying under 
SM, also computed from the origin. The cumulated SM-LM difference 
to the right of OA must be no smaller than the cumulated LM-SM dif- 
ference to the left. 3 In the graph this is supposed to occur at output OB. 
In other words, as long as the old equipment is available, the marginal 
cost curve which is relevant for decisions to expand or to contract will 
possess a discontinuity. When the MR curve shifts contraction will occur 
along the SM curve; and expansion will occur also along the SM curve up 
to the output marked OB in Figure 2. At this output the aggregate total 
cost with the new plant 4 ceases to be greater (becomes smaller) than the 
aggregate variable cost with the old. 5 Hence from output OB to the right 
the LM curve becomes relevant. The relevant marginal cost curve is SM 
up to C in Figure 2, at which point it drops discontinuously to point D 
and then continues to the right along LM. 

Assume now that the MR curve of Figure 2 shifts to the right. If the 
shifted curve passes through a point due north of C, then the relevant 
intersection is that of the shifted MR curve with LM. If the shifted MR 
curve passes through a point due south of D, the relevant intersection is 
that with SM. The critical degree of shift is that which takes MR through 
the C-D gap in such a way that two triangles should be equal, namely, 
the triangle bordered by MR, SM, and that part of the C-D stretch lying 
above MR (upper-left triangle), and the triangle bordered by MR, LM, 
and that part of the C-D stretch lying below MR (lower-right triangle). 
If these two triangles are equal, the firm is indifferent as between the 
MR-SM intersection and the MR-LM intersection. If the shift of MR 
exceeds this critical degree, and thus the lower-right triangle becomes 
greater than the upper-left, then the MR-LM intersection becomes 
relevant, and the expansion occurs along the long-run curve. The ex- 
planation of this proposition is that for the output OB profits would be 
the same on the short-run as on the long-run curve; as compared to these 
profits a profit increment is obtainable which is measured by the upper- 
left triangle if the firm moves from OB to the intersection of the shifted 
MR curve with SM, and is measured by the lower- right triangle if the 

3 We have disregarded the scrap value of the old equipment. This should be 
added to the right-hand difference or subtracted from the left-hand difference. 

4 Therefore also the average total cost with the new plant. 

5 Therefore also than the average variable cost with the old plant. 


firm moves to the intersection with LM. I would like to express my debt 
to Mr. Alvin Marty, who called my attention to the fact that my earlier 
conception of the minimum condition was faulty, in that my test was in- 
sufficiently severe. 

The paragraph which now follows contains an adjustment of the pre- 
ceding propositions to conditions where expansion along LM requires 
merely partial scrapping of the old plant. The preceding presentation of 
the matter assumes that a firm expanding along LM must completely 
scrap the existing plant. If this is not so, the presentation must be changed 
but, except for a limiting case, the same kind of conclusion emerges. Ex- 
pansion along LM may be associated with less than total scrapping: The 
divisibilities of the existing plant may be such that the larger optimum 
plant corresponding to larger outputs consists partly of the same elements 
as the existing plant which was built for output OA. In this event the 
C-D gap will occur at a smaller output than that at which it is made to 
occur in Figure 2. For in Figure 2 it occurs where the cumulated dif- 
ference between SM and LM to the right of OA becomes equal to the 
cumulated LM-SM difference to the left of OA. But if expansion along 
LM requires only partial scrapping of the existing plant, then the SM-LM 
difference to the right of OA will have to be made equal to less than the 
cumulated LM-SM difference to the left. The reason for this is that even 
after enlargement of the plant, the firm can produce any output up to 
OA less expensively than the LM curve indicates because it produces 
these outputs partly with existing (old) plant elements. Consequently 
we must draw a curve between SM and LM in the left-hand section of 
Figure 2. Merely the cumulated difference between this broken curve 
and SM, to the left of OA, must be made up by the cumulated dif- 
ference between SM and LM to the right. At the point where this con- 
dition is satisfied, the C-D gap occurs. That is to say the firm will expand 
along SM up to this point, and along LM beyond this point. The relevant 
marginal cost function is SM from the origin to this point and LM from 
this point on. In the graph the C-D gap has not been redrawn but it is 
obvious that the gap corresponding to the broken curve should lie to the 
left of the original gap. The limiting case is that of complete divisibility 
owing to which an optimum plant corresponding to higher output always 
consists of an optimum plant corresponding to lower output plus a 
distinct (separate) addition. In this limiting case the broken curve ceases 
to lie betvceen LM and SM. Instead, it coincides with SM up to OA. In 
this limiting case there exists no cumulated left-side difference which 
must be made up to the right of OA. The C-D gap disappears: all ex- 
pansion beyond OA will occur along LM, and of course all contraction 
below OA continues to occur along SM. 6 

6 This limiting case can best be visualized as one in which diseconomies along 
the cost curves of single "plants" in the conventional sense (i.e., of single con- 
stituents of the total plant of the firm) become significant at a negligible fraction 



Let us now for a moment forget about the "broken curve" of the 
preceding paragraph, that is to say, let us assume that expansion along 
LM requires total scrapping of the existing plant. A useful invention 
lowers LM by definition but, of course, it does not change the SM curve 
which must be used for comparing the new method with the old. There- 
fore the innovation increases the C-D gap. It also shifts this gap to the 
left because along the lower LM curve it takes less additional output (in 
excess of OA) to make the area under the long-run curve smaller than 
the area under SM. However, to induce the firm to scrap its plant or to 
lower its price, the widening and leftward shift of the gap must be great 

Figure 3 

enough to have the MR curve either pass through a point due north of 
C, or have it pass through the C-D gap in such a way that the lower- 
right triangle described previously should be greater than the upper-left 
triangle. If this condition is met with the LM (new) curve, we shall say 
that the invention passes the Figure 3 test. The invention will be 
promptly adopted as an industrial innovation if and only if it passes the 
Figure 3 test. 

Here again a qualification is required, if part of the old plant can be 
employed for production with the new method. The following paragraph 
contains the required addition which must obviously be analogous to that 
explained in connection with Figure 2. Turning to Figure 3, we have to 
draw a curve between the original SM and LM (new) to the left of the 
intersection of these two curves and ending in their intersection point 

of the firm output. This in turn assumes that "interplant diseconomies" should not 
stand in the way of expanding the output of the finn to an exceedingly high mul- 
tiple of the output of the single "plants." Cf. Don Patinkin, "Multiple-plant Firms, 
Cartels and Imperfect Competition," The Quarterly Journal of Economics, Vol. 
LXI (February, 1947). 


(P)J This hybrid curve — intermediate between long and short run — 
will again be "inserted" as a broken curve. In its unrefined form (assum- 
ing no overlap between the two types of plant), the Figure 3 test does 
not call for adding such a curve because the unrefined test requires that 
the cumulated difference between the original SM and LM (new) to the 
right of P should be no smaller than the cumulated LM (new) — SM 
(original) difference to the left of this point. Merely these two curves 
are needed for expressing the "unrefined" condition. However, now we 
have to add that, in the statement of this condition, the cumulated dif- 
ference between LM (new) and SM (original) to the left of P should be 
replaced by the cumulated difference between the broken curve and the 
original SM curve. This is the amendment which becomes necessary if 
elements of the existing plant can be employed for the new technique. 
For, if the new technique is used, and if this implies merely partial 
scrapping of the old plant, then the firm is free to move along the 
broken curve, rather than along LM (new), up to the intersection of LM 
(new) with the original SM. In the entire region to the right of this 
intersection the firm will prefer to move along LM (new) and in this 
region it will gain, by adopting the innovation, the cumulative difference 
between the LM (new) and the original SM curve. To the left of the 
intersection it loses, by adopting the innovation, the cumulated difference 
between the broken curve and the original SM. Hence in the refined 
version of the Figure 3 test, the C-D gap (which was not redrawn) 
should occur where the cumulated difference between the original SM 
and LM (new), to the right of P, becomes equal to the cumulated dif- 
ference between the broken curve and the original SM curve to the left 
of P. Passing the test requires that MR should pass through the gap in the 
way already described or should lie even higher. 

The gap problem disappears, and the innovation becomes promptly 
adopted, regardless of the degree of cost saving, in the limiting case 
where no scrapping is required to put it into effect. This means that the 
SM curve itself shifts down and that the broken curve coincides with 
the new SM. In no range of possible outputs does the introduction of the 
innovation cause a loss because this curve is lower than the original SM 
which remains valid for the comparison. 

Moreover, it should be pointed out that "in the long run" (but pos- 
sibly with a significant time-lag) the test practically always becomes 
satisfied because the gap — which usually does exist — tends to shift to the 
left. The technological reason for this is that current "piecemeal" physi- 
cal maintenance practically always is less than complete. The SM curve 
shifts up as the physical condition of the old equipment deteriorates. 
The SM-LM intersection in Figure 2, and point P in Figure 3, gradually 

7 The broken curve is a hybrid marginal cost curve. It is lower than LM 
(new) by the fixed cost corresponding to that part of the old plant which can be 
used as a constituent of the new plant. 


shift to the left, and the cumulated differences must now be computed to 
the right and to the left of these shifted intersection points. The gap oc- 
curs where the right-hand difference ceases to be smaller than the left- 
hand difference. The leftward shift of the gap in Figure 2 leads to re- 
placement of the old equipment, even if no inventions are made. In some 
(possibly distant) future period even an unchanging MR curve will go 
through the shifted gap in Figure 2 in the way required by the test. 8 
Furthermore, in some future period, the MR curve of Figure 3 will go 
through the shifted gap in that figure even if "in the short run" point D 
lies to the right of MR. At that time the test will be passed even if 
originally the invention has jailed it. 

Finally it should be noted that the MR curve of the period in which a 
decision is made concerning innovations need not pass through the LM 
(o\d)-SM intersection. The existing plant was designed on the expecta- 
tion that the MR curve would pass through this intersection as shown 
in Figures 2 and 3. But the MR curve may have turned out differently. 
If it has shifted to the right, as compared to its originally assumed equilib- 
rium position, this increases the chances that a small improvement will 
promptly be adopted. A leftward shift decreases these chances. A shift 
to the right may have brought the MR curve very near to the C-D gap 
in Figure 2, before the innovation. Then a comparatively small improve- 
ment and the attending small widening and leftward shift of the C-D 
gap will make the MR curve pass through the gap. 

Failure of firms promptly to adopt improvements which do not pass 
the Figure 3 test will here be called simple retardation. This kind of re- 
tardation and also the kind which in Section III will be called anticipa- 
tory retardation are consequences of fixed outlays or overhead costs. 9 
More specifically, they are consequences of overhead which (a) stems 
from jointness of some costs to which outputs of different periods give 
rise, and (b) is at least partly specialized to a definite method of produc- 
tion. Innovations will be promptly adopted whenever the overhead is 
not specialized so that all plant requirements of the two methods are 
identical; or where the cost-saving overrides the specialized overhead. 
At this point it should be remembered that existing firms practice 
simple retardation even if entry is free. Even the firms of the purely 
competitive industry portrayed in Figure 1 practice simple retardation. 
However, the degree of retardation which is practiced by the existing 
firms may depend on freedom of entry because pre-invention entry may 
make it easier or more difficult for an invention to pass the Figure 3 
test. 10 Unexpected pre-invention entry shifts the MR curves to the left, 

8 This seems to provide an adequate framework for replacement analysis. How- 
ever, allowance must be made for scrap values. Cf. footnote 3. 

9 Cf. J. M. Clark, Studies in the Economics of Overhead Costs (Chicago, 1923), 
mainly pp. 191-92. 

10 Post-invention entry comes too late to affect the test for the invention in 


out of the SM-LM intersection point, and this makes the test more diffi- 
cult. Correctly anticipated pre-invention entry makes the intersection 
points as well as MR lie more to the left and it need not affect the ease 
of the test. In fact, it will not affect the ease of the test if the individual 
demand elasticities are independent of the number of firms and if the 
indivisibilities of equipment are the same for small as for big firms. If 
pre-invention entry has increased the individual demand elasticities, as 
may frequently be the case, it becomes easier for an invention to pass the 
test. On the other hand, if pre-invention entry and the reduction of plant 
size have made the individual plants less divisible, as may also frequently 
be the case, then it becomes more difficult for an invention to pass the 
test because for small plants the broken curve may disappear or it may 
run very close to the LM (new) curve. It is impossible to indicate with 
general validity whether entry increases or reduces the ease with which 
the test is passed. 

We have seen that it is possible to indicate the general nature of an- 
other effect of entry. If free entry exists — and only to the extent to 
which it exists — entrants will promptly adopt the new method 11 even if 
it does not pass the Figure 3 test, and they will force the old firms to 
adjust their prices to the level corresponding to the improved technique. 
This inflicts losses on the old firms which they would not suffer in a 
static economy. Nor would they suffer these losses in a dynamic econ- 
omy with blocked entry. If entry is blocked, price and output remain 
unchanged unless (or until) the invention passes the test expressed in 
Figure 3. If entry is free the price declines and aggregate output grows 
(progress occurs) immediately under the impact of any invention that 
lowers the long-run marginal cost. This is a generally predictable dy- 
namic consequence of entry. 

For logical completeness a section should now follow on a flow of 
unforeseen inventions. In the real world inventions come in flows, not as 
single events. But all that need be said here is that each unforeseen in- 
vention has the consequences discussed in the preceding sections and 
that these consequences depend on the freedom of entry in the same way 
as did the consequences of the single invention so far considered. 


It will now be assumed that firms gear their innovation policy to the 
expected flow of new inventions. Each invention is viewed as a link in 
a chain. In the present section it will also be assumed that anticipations 
are perfectly correct. 

Correctly anticipating firms will obviously practice retardation in the 
sense of not adopting each invention that lowers the long-run marginal 

11 In the event of monopoly or oligopoly based exclusively on real-cost advantages 
of scale, they may merely threaten to enter with the new method and this will 
have the same result. 


cost. Moreover, the anticipatory retardation here considered will be dif- 
ferent from the simple retardation defined previously. Assume two firms 
operating at a given initial moment with identical techniques. New in- 
ventions are made and one firm practices simple retardation (wrongly 
viewing each invention as a once-for-all change) while the other has 
correct anticipations. It is possible to state that during a time period, 
which follows the initial moment, simple retardation will be a weaker 
retarding factor than anticipatory retardation. Correctly anticipating 
firms will not adopt innovations which do not pass the test of Figure 3 
and hence are not adopted under simple retardation. But correct anticipa- 
tions may very well induce a firm not to adopt an innovation which 
passes the test of Figure 3 (and hence is adopted under simple retarda- 
tion) if this will shortly be followed by a superior innovation possibility 
requiring a different kind of equipment. If for this reason, in the first 
phase after the initial moment, more advanced methods are used under 
simple than under anticipatory retardation, then this phase may be fol- 
lowed by one in which anticipatory retardation goes with more ad- 
vanced methods. The correctly anticipating firm will now adopt the 
superior method for which it had been waiting while the firm practicing 
simple retardation may be "stuck" with the first innovation that passed 
the test of Figure 3. Once this "first innovation" is introduced, it may 
not pay to scrap the equipment and adopt the superior innovation for 
which the correctly anticipating firm was waiting. 12 But if in certain 
phases of development anticipatory retardation goes with more advanced 
techniques than simple retardation, then these must be preceded by phases 
of which the reverse is true. 

Anticipatory retardation produces the maximum available excess of dis- 
counted future revenues over discounted future costs, that is to say, it 
maximizes present net worth. An essential element of the procedure by 
which this is accomplished consists of by-passing inventions which will 
be followed by superior ones soon enough. It follows that under correct 
anticipations an invention will be adopted only if it passes a test which is 
more severe than that expressed in Figure 3. The invention must satisfy 
the Figure 3 test, and it must not be surpassed in the sufficiently near 
future by a sufficiently superior subsequent innovation. We shall express 
this by saying that the invention must pass the anticipatory test. 


Free entry brings reduction of firm size and possibly higher individual 
demand elasticity. Higher demand elasticity makes it easier for an in- 

12 This is a "may not" rather than a "cannot." The fact that the correctly an- 
ticipating firm finds it profitable to by-pass the first innovation does not prove 
that, in the event of the introduction of the first innovation, it is unprofitable to 
scrap and to adopt the superior one. This may still be profitable but less so than 
to by-pass the first. But it may be unprofitable to scrap, once the first innovation is 


vention to pass the test. The reduction of firm size may go with increas- 
ing individual plant indivisibilities, which make passing of the test more 
difficult. Hence free entry may change the ease with which an invention 
is capable of passing the anticipatory test. No general statement can be 
made on whether the freedom of entry increases or reduces the number 
of inventions that are capable of passing the test. This is analogous to 
what has been said previously about the effect of entry on the content 
of the Figure 3 test, except that here we do not have to allow for un- 
expected entry and need not distinguish between pre-invention and post- 
invention entry. 

As long as inventions were viewed as unforeseen single events, it was 
necessary to add that entry had a further and predictable effect on the 
rate of progress (i.e., on the rate of price reduction and output growth). 
Inventions that did not pass the test for the existing firms were adopted 
by intruding newcomers who inflicted losses on the old firms by forcing 
them to adjust their prices. This further effect of entry disappears if the 
expectations of all firms are correct. New firms, unhandicapped by over- 
head, will still enter with new methods even if the innovation does not 
pass the test for firms already existing, and new entries will produce a 
continuous price decline. But the existing firms are now assumed to 
know this in advance, and the entrants of any period are assumed to 
possess the same knowledge concerning the subsequent periods. This 
means that the consequences of all future entries are taken into account 
in the revenue expectations and output decisions of each firm. Firms will 
not enter if the expected revenue is insufficient for recovering the total 
cost and the firms that do operate will establish production plans which 
maximize the excess of discounted future revenue over cost, with due al- 
lowance for entries with new methods. If for a future period new entries 
must be taken for granted, then the number of firms other than netv en- 
trants (i.e., the number of previous entrants) will be correspondingly 
smaller so as to have the total number and the price come out right for a 
period properly chosen. We shall now examine the process which ac- 
complishes this result. 

In the event of unqualified freedom of entry into an industry (pure 
competition) it will be true of all firms that, from the time of the con- 
struction of a given plant to the time of its scrapping, precisely the sum 
total of costs will be recovered. In the beginning of each of these suc- 
cessive periods in the equipment history of a given firm, the price will 
be higher than the conventional ATC because entrants using new meth- 
ods are expected to force it down; toward the end of each period the 
price is lower than the conventional ATC because entrants actually have 
forced it down. 13 But for each firm the representative price (average 

13 The corresponding period of the equipment history of the successive entrants 
starts of course in the successive moments in which they enter with improved meth- 


price), 14 for the construction-to-scrapping period taken as a unit, will 
have to equal ATC. This means that with correct expectations un- 
qualified free entry merely has the effect of producing long-run equal- 
ity of price with ATC (i.e., of preventing an excess of price over ATC), 
which of course is a basic proposition of static price analysis. A specifi- 
cally dynamic effect develops only if entry changes the length of time 
elapsing between construction and scrapping, i.e., affects the test which 
an invention must pass to be adopted by already existing firms. If it 
fails to do this, the same basic periods (from construction to scrapping) 
emerge under free as under blocked entry, and these basic periods follow 
each other at a rate unaffected by entry. 

For the sake of illustration, we may refer to the purely competitive in- 
dustry pictured in Figure 1. Take a unit period, starting when plants are 
built along cost curve LS and ending when these plants are scrapped. 
Scrapping will not take place before inventions have lowered LS and 
ATC far enough to make the new LS-ATC tangency lie below the AVC 
curve of the old plants. These A VC curves may be conceived as remain- 
ing unchanged, or, more precisely, as being gradually shifted up by the 
wear and tear of the old plants. Our unit period will not end before the 
new LS-ATC tangency sinks below AVC because this is also the time 
when the Figure 3 test becomes satisfied for the firms that have built 
their plants along the original LS curve. In this case the MR curve of 
Figure 3 is horizontal, and both cost curves are upward sloping. Con- 
sidering the nature of the anticipatory test, scrapping may occur later 
than the time when the Figure 3 test becomes just satisfied, provided 
that sufficiently superior further inventions will follow soon enough. 
Our unit period ends when these plants are scrapped and replaced by 
superior ones. In the beginning of this unit period the price will be in 
excess of OL and at the end of the period it will be below OL. The lower- 
ing of the price will be brought about by new firms which have no 
initial overhead. However, the "average" or "representative" price of the 
period will be OL. This statement takes into account the fact that when- 
ever the cost is lowered new entrants, with no previous overhead, must 
be assumed to lower the price; and it also takes account of the fact 
that for a period properly chosen total cost must be recovered. The en- 
tire future history of the industry may be described as a continuous 
succession of construction-to-scrapping periods of this sort, regardless 
of where we start. 15 In each period the initial total cost will be recovered. 
The clock-time duration of the successive basic periods need not of 
course remain constant. 

14 This is a weighted average, due to the discounting. The weight of the far- 
future is smaller than that of the near-future. 

15 For example, we could have started with firms that entered during our "pe- 
riod" and could have called the moment of their entry the beginning of a unit 
period. These firms too will recover precisely their total cost (which, however, 
will be lower) . 


The essential difference between this outcome and the outcome in 
Section I (unexpected innovation, free entry) is that in Section I it was 
impossible to define a period for which price and output would be the 
same as under static equilibrium conditions, and then to argue that, due 
to improvement, one such period is succeeded by another (higher-out- 
put period) at the same rate under free as under blocked entry. It is true 
that blocked entry will not prevent old firms from eventually adopting 
an innovation, regardless of whether expectations are correct or wrong 
(because "in the long run all costs are variable"). But this does not mean 
that in Section I (incorrect expectations) the long-run dynamics of the 
process were unaffected by entry. On the assumption of Section I, free 
entry led to a price decline and to the growth of output earlier than 
would have been the case under blocked entry (cf. Section II), and there 
existed no subsequent offset to this. A further flow of unexpected in- 
ventions would clearly not create a subsequent offset. It would merely 
result in further effects of the same sort. On the assumptions of the 
present section (correct foresight) the price is higher than OL, and the 
output correspondingly lower, in the beginning of each unit period but 
this is offset by the reverse in the later course of the same period, and 
with due allowance for time preference (discounting). The rate of 
growth of output from unit period to unit period — the long-run rate 
of growth — is uninfluenced by entry. This is what we mean by saying 
that entry does not influence the long-run dynamics of the process. Entry 
has "static" consequences because in the absence of entry the price for 
each unit period would exceed OL; it also affects the short-run dynamics 
of the process, by producing price movements during each unit period. 

This conclusion assumes correct expectations. It also assumes that 
entry does not affect the demand elasticity for the products of individual 
firms, or their plant indivisibilities, in such a way as to make it easier or 
more difficult for a given invention to "pass the test" from the viewpoint 
of existing firms. If entry changes the content of the test, then the unit 
periods to which essentially "static" analysis applies succeed each other 
at different rates depending on whether entry is or is not free. But if 
these two assumptions are made (correct foresight and no change in the 
content of the test), then the long-run consequences of entry are merely 
those established by static analysis. There exist no additional long-run 
consequences. At each level of technological knowledge price will be 
lower and output higher when entry is free but the long-run rate of 
progress (or of output growth) will be the same. Professor Schumpeter 
expressed a similar conclusion by suggesting that the monopolistic rate of 
progress tends to be the same as that developing under an ideal form of 
"planning." 16 Given the assumptions which were repeated in the present 
paragraph, we arrive at essentially the same conclusion. 

16 Cf. Joseph A. Schumpeter, Capitalism, Socialism, and Democracy (2d ed.; New 
York, 1947), chap. viii. 



On the static level of analysis it is possible to conclude from widely 
accepted welfare postulates that institutional barriers to the entry of 
producers are harmful. The minimum axiomatic requirement for ac- 
ceptance of this conclusion is that a movement toward equality of price 
and marginal cost (and more generally toward equality of factor price 
and value of marginal product) is desirable provided the accompanying 
change in the distribution of income is not considered undesirable. The 
change in income distribution which accompanies the elimination of 
artificial barriers is a reduction of quasi-rents and of haphazard subsidies 
to the consumers of the relatively unrestricted industries. This is usually 
considered a desirable change. 

The preceding sections lead to the conclusion that, as long as profit 
is maximized on correct anticipations and "passing of the test" is made 
no easier, there exists no additional dynamic argument against barriers to 
entry. In pure competition the profit-maximizing price equals marginal 
cost, while when entry is blocked the profit-maximizing price exceeds 
marginal cost. This is found to be true on the static level and the cor- 
responding difference shows also when profit maximization occurs in 
correct anticipation of future changes. But, aside from the possible effect 
of firm size on the test, we have found no difference in the long-run 
rate of growth of output between closed and open groups. At each level 
of the arts output is affected by the degree of competition, but stating 
this merely repeats the results of static analysis. 

With profit maximization based on correct anticipation, the welfare 
case against restriction would have to rest on its static merits, or on 
the very doubtful notion that entry generally makes it easier to "pass the 
test." The static merits of the case should not be overlooked but in the 
actual world a purely static argument is of limited significance. Data 
computed by American government agencies point to an increase in 
man-hour output of between 25 per cent and 33 per cent for each decade 
from 1890 to 1950, with the average increase per decade approximating 
30 per cent. Man-hour output would not remain constant even on a con- 
stant level of technological knowledge, but it is likely that technological 
progress is an important cause of the observed increase. Unless a very 
short view is taken of social processes, one will be inclined to attribute 
more significance to a moderate acceleration or retardation of the rate 
of progress than to unchanging degrees of malallocation or maldistribu- 
tion on given levels of technique, within the range in which such malal- 
location is likely to occur at all. The static approach is highly incomplete. 

If profit maximization based on correct anticipation is postulated, the 
difference between blocked and open entry is merely that discernible 
by static analysis. The static difference establishes a general presumption 
against barriers. However our ultimate judgment should not be based 


on this presumption alone but also and perhaps primarily on dynamic 
differences caused by incorrect anticipations \ and also by deviations from 
profit maximization such as are due to nonprofit objectives and inertia. 
With incorrect anticipations and with deviations from profit maximiza- 
tion, important differences may develop between the rate of progress 
under free and blocked entry, or under "competition" and "monopoly" 
respectively. The analysis of Sections I and II illustrates this. These two 
sections assumed a specific kind of wrong anticipations. 

It seems to us that the unsatisfactory character of the discussion about 
"workable competition" is a consequence of the fact that it is necessary 
to rely on the record of specific industries to obtain indications concern- 
ing the role of factors such as excessive optimism or pessimism, inertia, 
etc. In specific industries with atomistic competition and free entry the 
future rate of progress may be underestimated, and hence the kind of loss 
suffered by the firms of our Section I may grow to substantial size. This 
goes with a rapid rate of expansion but it may give rise to intervention 
on grounds of equity or owing to repercussions on the credit system or 
on other industries. Alternatively, strongly dynamic periods in the his- 
tory of atomistic industries may be succeeded by long periods of ex- 
cessive entrepreneurial caution and of slow progress. This is undesirable. 
On the other hand, monopolists, even if they should have a more reliable 
appraisal of the future rate of invention, may not always satisfy the re- 
quirements of profit maximization because, with their profits at a gen- 
erally satisfactory level, other objectives (freedom from worry, prestige, 
popularity) tend to acquire added marginal significance. This too may 
have undesirable consequences. The influence of market structure on the 
rate of progress depends primarily on factors of this sort. 

The same matters bear importantly on the oligopoly problem. Typi- 
cally, oligopolistic coordination of business policies (so-called collusion) 
is less complete for technological improvement, including quality im- 
provement, than for price policy. 17 Is this incompleteness of "collusion" 
desirable? It leads to more rapid progress than does complete collusion 
because each firm is trying to gain at the expense of its competitors. If 
the effects of such behavior were correctly foreseen and the behavior 
was nevertheless not abandoned, the consequences of the increased rapid- 
ity of progress would in some sense be offset by higher prices. The 
problem is analogous to that of free versus blocked entry. The tighter 
the collusion, the less free is each firm to "enter" into the market area 
of the others. 

However, in this case it is obvious that some firms underestimate the 
innovating ability of their competitors as compared to their own ability, 
and that this blocks the kind of cooperation by which the joint profit 

17 The present writer discussed this aspect of the problem at some length in his 
Competition Among the Few (New York, 1949), mainly on pp. 33-40, 183-90, 


could be truly maximized. Some firms underestimate the amount of 
"entry" (intrusion) into their markets of which others are capable, and 
others are forced to accept the competitive policy which is imposed on 
the group by the overoptimistic. Correct foresight would result not in 
quality competition plus offsetting price increases but in collusive han- 
dling of improvement. For a completed period it is always possible to 
describe a collusive pattern with respect to quality and price which 
would have given each participant the relative share he actually happened 
to obtain and which at the same time would have given each participant 
higher absolute profits. Incomplete collusion is incompatible with per- 
fect anticipations. It expresses willingness to act on the assumption that 
the other firm's competitive skill in relation to one's own is smaller than 
the other firm's estimate of it. Hence the consequences for the group as 
a whole are very likely to be the same as those of overoptimistic ex- 
pectations of existing firms concerning the inability of outsiders to enter. 
As compared to its own guess of relative skills each firm may proceed 
cautiously and conservatively. But oligopolistic quality competition in- 
dicates that the guesses of some firms are overoptimistic, and this sets 
the pace for the group as a whole. The consequences of such limited 
overoptimism are socially "desirable" provided that it is felt to be desir- 
able to obtain more improvement coupled with a distribution effect ex- 
pressing itself in the reduction of monopolistic or oligopolistic incomes. 
Furthermore, oligopolistic cooperation concerning the introduction of 
improvements is likely to be the less complete, the greater the number 
of firms included in the oligopoly. Therefore, if the consequences of 
reduced "collusion" are considered desirable, then broadening of oligop- 
olistic market structures should also be so considered. Here again the 
axiomatic requirements for a favorable welfare judgment are com- 
paratively mild. 18 

To repeat: with correct foresight and complete rationality, dynamic 
analysis of the monopoly-oligopoly-competition problem, or of the 
problem of free versus blocked entry, discloses no essential modification 
of the static results. The specifically dynamic aspect of the problem calls 
primarily for weighing, in specific instances, the probable consequences 
of incorrect foresight, and also the effects of deviations from profit 
maximization due to inertia, high safety preference, noneconomic ob- 
jectives, etc. This points to the necessity of making specific judgments 
in specific cases. Yet it is possible to distinguish broad classes of instances 
in which a strong presumption exists in the one or the other direction. 
Entry, the effects of which on the rate of progress are less than fully off- 

18 It is true that many persons feel skeptical toward some of the "quality im- 
provement" developing in oligopolistic markets. But this merely illustrates the fact 
that even comparatively mild general welfare axioms should be applied with ad hoc 


set by anticipatory price charges, is desirable unless the resulting change 
in income distribution is regarded as undesirable. The behavior of mo- 
nopolistic and oligopolistic groups suggests that such is typically the out- 
come when narrow market structures are broadened. The case for broad- 
ening narrow market structures remains fairly generally valid on the 
dynamic as well as on the static level of economic analysis, even though 
on the dynamic level no general case emerges for atomistic competition. 
The case also remains valid from a general sociological point of view for 
any society that considers freedom of opportunity a desirable objective. 


So far we have assumed that each new method becomes immediately 
available to any interested firm. The conclusions of the last section must 
be supplemented by a brief discussion of limited availability. We shall go 
but little beyond calling attention to the main consequences of this 
specific "barrier" which is strengthened by patent legislation but would, 
to some extent, exist even without legislative action. Know-how, when 
it is new, begins to spread gradually. It never becomes completely dif- 

These would seem to be the most significant consequences of limited 

(1) Inventive activity is stimulated because the reward for making 
inventions is increased. Some argue that this effect is frequently over- 
estimated for the present environment in which the individual (or 
independent) inventor plays a smaller role than the corporate research 
organization. Quantitative appraisal is impossible but it is difficult to im- 
agine that inventive activity should not to some extent be stimulated by 
limited availability. 

(2) The introduction of "inventions" as industrial "innovations" is also 
stimulated because no technological invention is really and truly made 
before it has been proved to work on an industrial scale. In other words, 
the conventional distinction between invention and industrial use, which 
was implied under (1), is sufficiently arbitrary to call for the qualifica- 
tion which is here added. No firm ever knows with certainty that a 
process is going to work before this has been demonstrated industrially. 
Innovating is part of the process of inventing. 

(3) Once the process is introduced, limited availability is a "barrier" 
or a limitation of entry and hence it has the further consequences with 
which the main sections of the present article were concerned. These are 
the traditionally emphasized static "harmful" consequences of "monop- 
oly" and the specifically dynamic consequences discussed in Section V. 
The dynamic consequences of restriction are presumably more signifi- 
cant. They stem mainly from the incorrectness of expectations and, 
generally speaking, from the incompleteness of profit maximization. Un- 


like the static consequences they call for specific appraisal of individual 
instances or of certain "classes" of instances. In some cases the dynamic 
consequences may be found "undesirable," in others "desirable." 

(4) Limited availability affects the size of the firm and thereby pre- 
sumably also the rate at which further inventions are made. The relation- 
ship between the size of the firm and efficiency of its research organiza- 
tion gives rise to significant questions about which not much seems to 
be known. The optimum firm size is not truly optimum if its research 
activities are sufficiently far removed from optimum efficiency. If the 
relationship between firm size and research efficiency is unknown, both 
the automatic market mechanism and policy are likely to favor an un- 
duly static kind of "optimum." 19 

Considerations based on (1) and (2) lead to a case for patent legisla- 
tion, or more generally, for limited availability. In some specific instances 
the case for restriction may become reinforced by the "dynamic" con- 
siderations included in (3). Misgivings against patent restrictions are 
usually based on the "static" considerations in (3) and on the notion that 
the "dynamic" considerations in (3) may frequently reinforce the case 
against restriction. It is usually recognized that considerations based on 
(4), taken in isolation, do not favor atomistic competition but at present 
it is impossible to form an opinion of what firm size they do favor in 
various industries. 

19 Marshall, in Book VI, chap. 7 of his Principles, places considerable emphasis 
on the static bias of market selection (i.e., of success in the market) . This ties in 
with the problem discussed in the text. 


Toward a Theory of Industrial 
Markets and Prices* 


Generalizing about markets and price formation in the industrial sector, 
where fewness of participants and differentiation are usual attributes, is 
a complex task which requires one to draw on price theory, on the em- 
pirical research of the past twenty years, and on his own observations 
and hunches. Few parts of the conclusions reached are beyond the hy- 
pothesis stage. When, however, hypotheses about each of the various 
facets of the problem are woven together, the outline of a theory emerges 
which appears useful in explaining the operation of industrial markets 
and in prognosticating their behavior. 

Theory in this area must be directed toward exactly the same problems 
as those to which neoclassical price theory has been addressed, but ex- 
planation of industrial markets proceeding inductively from empirical 
evidence is often held to higher standards than is the deductive theory 
of prices. Both approaches must identify the variables that are relevant 
for a given problem. Both must demonstrate the net results of the inter- 
action of these variables under various assumptions as to their magnitudes 
and the extent to which these results would accord with established tests 
for optimal use of resources. Price theory, however, stops with this and 
in so doing serves essentially as a computing machine into which data can 
be put and conclusions ground out. But those generalizing in the area 
under discussion are usually expected to indicate the actual empirical 
content of the variables as far as that is possible. 


Several requirements for an adequate theory of industrial markets and 
prices can be set down. Even for a static theory, it is not possible to build 
the analysis on given utility and technological functions, as is done for 
competitive and monopolistic price theory. 1 To varying degrees, the 

* The American Economic Review, Vol. XLIV (Proceedings of the American 
Economic Association, 1954), pp. 121-39. Reprinted by courtesy of the publisher and 
the author. 

t Northwestern University. 

1 This point has been most fully expounded by William Fellner in his Competi- 
tion Among the Few (Knopf, 1949), pp. 8-13. 



demand and supply functions in markets that lie in the zone intermediate 
between the structurally competitive and the monopolistic are themselves 
results of market processes and must be explained. Added to the diffi- 
culties of static theorizing in this area is the need for dynamic analysis 
and for broadening the variables considered — all of which points to the 
following requirements for a satisfactory theory: 

1. A satisfactory theory must be historically related in the sense that 
technological, organizational, and public policy influences of a certain 
economy at a certain period are assumed. Here we refer to the present- 
day American economy, where the public policy is designed to keep 
cartel arrangements so weak that group control over important market 
variables is not significant for long. 

2. Ideally, the theory should be able to explain why the market is as it 
is, because structure does not necessarily emanate from technological and 
utility functions. 

3. A satisfactory theory must often be dynamic because market struc- 
ture is potentially variable and because the technology actually used 
and many other of the diverse forms of interfirm rivalry are optional. 
Those selected may set in motion a chain of reactions throughout the 

4. A satisfactory theory must go beyond explaining the cost-price-out- 
put relations for a given product and must be prepared to appraise the 
conditions of choice among other variables and the economic results of 
that choice. This requirement, like the others set forth here, is not 
introduced for the sake of descriptive realism but because it is essential 
for better prediction. 

One general feature of all of these requirements is that in the search 
for uniformities one must not abstract from diversities that are relevant 
to the behavior to be explained. Whether narrow- or broad-gauge theory 
is sought, it is obvious that definitive theory which meets the above 
specifications cannot be laid down at this time. 


An advance sketch of the general line of argument and some of the 
assumptions on which it is based will place each of the later sections in 
better focus. 

1. This is not an attempt to theorize about firms in general but rather 
about markets and collections of firms that constitute industries which 
have certain characteristics. (The delimitation of an industry or a market 
will always be relative to the context in which these terms appear.) 

2. The industry or market is looked at in midstream. Its structure is 
not traced back to ultimates or examined as if it were being established 
de novo. 

3. The emphasis is on the various factors that influence the structure 
of the market and on the effect of the structure in turn on the market's 


operation. Relevant, therefore, are such basic conditions as the nature of 
final buyer demand and the technology available for production. But final 
buyer demand, as it bears on the manufacturer, may be altered by the 
organization of the intermediate handlers or processors. Various sellers 
undertake to satisfy different, or several different, segments of the spec- 
trum of consumer wants in a particular area, or to deal in a different 
fashion with the distributive trades, or even to take on one or more as- 
pects of the distributive process. Within the same industry there are 
various degrees to which the whole manufacturing process is carried out 
by an integrated ownership rather than by several ownerships coordi- 
nated through purchase and sale. 

4. Both because of the ubiquity of diversity within markets and be- 
cause of the analytical usefulness of that characteristic, we shall first 
consider the apparently complex cases in which diversity among firms 
plays an important role. Cases in which such diversity is not significant 
will be treated secondarily. 

5. Markets for most products (broadly conceived) tend to settle into 
segments among which there are varying degrees of elasticity of sub- 
stitution and of intersegment mobility. Often the same firm will be doing 
business in more than one segment. 

6. Much of the analysis — probably the most important parts of it — 
should be directed toward the composite of industry segments, and for 
this we use the term "product line." 

The general argument of the following sections, which reflects the 
views just listed, is divided into the following parts: 

First, it is explained that by a historical process a state of balance is 
achieved in industrial markets and this balance 2 is not easily disturbed. 
Both the process of reaching a balance and its stability result from the 
diversity among firms each of which settles into one or more places in 
the loosely conceived market. Second, the economic attributes of this 
balance — the degree to which it is optimal — are abstracted from actual 
(but not from potential) variations in factor prices and industry demand. 
Economic results are found to emanate from entry and from the various 
forms of rivalry, choice among which reflects the structure of the market 
and the presence of uncertainty. Third, the degree and method of adapta- 
tion to factor price and demand movements are considered next with 
particular reference to whether external developments threaten to dis- 
turb the internal balance. Finally, an appraisal of the forms and prompt- 
ness of adaptation and of the economic character of states of balance in 
industrial markets is made insofar as available evidence and space permit. 

Because of limitations of space, no attempt will be made to show how 
the ideas advanced incorporate or differ from either deductive price 

2 Parenthetically, it should be observed that the term "equilibrium" is not used 
very often in the following pages even though by "balance" is meant a situation 
such that no one who has the power to change the situation wants to do so. 


theory or observations which have been made on the basis of empirical 
research. Few references will be made to the literature, much of which, 
it can be observed, stresses the noncompetitive performance of indus- 
trial markets. This omission gives an appearance of bias to the empirical 
evidence cited, but that risk must be run. 


Economic analysis depends heavily on discovering equilibrium tend- 
encies which reflect a sufficient balance of forces that economic proc- 
esses become orderly. The literature on markets where sellers are few 
and particularly where, in addition, a high proportion of costs are fixed 
in the short run, has been concerned with whether equilibria can either 
be determinate or be stable in the absence of collusion. Yet one observes 
that few such markets seem disorderly in the absence of evidence of 
collusion — a fact which is explained by the following arguments. 

The Process by Which Balance Is Established 

The structure of a market at a given time reflects an evolutionary proc- 
ess whereby firms come to acquire a workable relationship with one 
another. They are assisted in this process by the fact that most markets 
are made up of niches. The respective firms find one or more places for 
themselves by design or luck. 

It is in connection with the adaptation of the firm to its environment 
that uniquely farseeing or unwise business decisions have their major in- 
fluence. Once such decisions have been made (whether by accident or 
careful design), they set for some time the conditions under which 
shorter-term operating decisions will be made. Definitive conclusions are 
not possible as to the extent to which firms, autonomously, shape the 
varied attributes of the market structure, short, of course, of cases of 
market dominance. My own inclination is toward stressing the firm as 
consciously adapting to its environment or as being "adopted" 3 when its 
successful moves are accidentally good. 

Regardless of one's views on this controversial subject, it seems evident 
that a balance can be achieved among such forces as the desire for variety 
within a product area and the attendant services, the roles of various 
firms in supplying them, and relative prices. Such a balance is more 
multivariate than is one of prices and outputs. 

The relative importance of these variables differs with the group of 
products handled by the industry, but the historical tendency has been 
for the diversity of feasible roles for various participants in a market to 
increase. This has been true not only of consumer goods industries, as 
incomes have risen, but even of semimanufactured goods. Today finished 
goods manufacturers rarely start with crude materials but use semimanu- 

3 A. A. Alchian's terminology from his "Uncertainty, Evolution, and Economic 
Theory," Journal of Political Economy, June, 1950, p. 214. 


factured materials previously processed to fit their methods of manu- 

The evolutionary process does not work toward a balance in the pure 
oligopoly case, particularly where there are the added features of a low 
ratio of short-term marginal costs to total costs and a lack of oppor- 
tunity to use product quality as a variable. In such cases, either overt 
collusion or effective, tacit price leadership is apt to develop or the 
firms will integrate vertically into a stage in which the diversity required 
for stability does exist. In most markets, however, a balance is feasible 
without collusion, and, in such a balance, margins between direct costs 
and prices differ among sellers, among brands of or channels of sale by 
the same seller, and among physical variations of the product. 

The Stability of the Balance 

The characteristics of the environment which shape the market's evo- 
lution also govern its stability. The fact that firms occupy somewhat dif- 
ferent places in the market means that not all of them are in complete 
and direct competition with each other. But it also means that they can- 
not ignore or isolate themselves from one another, for their products are 
fairly close substitutes for each other and the analogous character of their 
production and selling operations facilitates entry into each others' back- 
yards. There remains, however, a zone within which they need not fear 
upsetting action by rivals. 

One reason for this is that at a given time each firm's freedom of 
action is limited by its history. The degree to which this is true depends 
on the adaptability of equipment, of personnel, of internal organization, 
and of relations with suppliers and customers. Usually, these inhibit 
quick, sizable changes in the character of the business. These funda- 
mental characteristics of the firm also reflect its longer-term goals, and 
temporary deviation from these goals is usually unwise. 

This last point leads me to emphasize that each participant acquires 
not merely a place in the market but also a position which is a major 
long-term attribute. By market position is meant more than what the 
firm sells and to whom and by what means, and more than its share of 
the volume. Instead, market position is that composite of attributes 
which governs the ability of the firm to compete. Obviously, the relative 
importance of elements in this composite differs among product areas and 
even among firms participating in the same product area. The relevant 
attributes of competitive strength may also vary according to the prob- 
lem. A vertically integrated firm, for example, may be strong for deal- 
ing with one type of problem and weak for another. 

Business executives are cognizant of their respective firms' market 
positions and of what that enables them to do and of what acts are un- 
wise. They realize that a strong market position is built slowly but can 
deteriorate rapidly. Above all, management recognizes good market posi- 


tion to be a valuable asset, whose long-term attributes must condition 
all short-term decisions. Its value is not merely defensive, as emphasis 
on security motivation suggests, but also is a basic attribute of the firm's 
ability to make positive moves; that is, to deal with unanticipated de- 
velopments when they occur. 

In that sense, market position becomes a means of long-term profit 
maximization under conditions of uncertainty. Consequently, the con- 
cept of market position is at the heart of the problem of the stability of 
industrial markets under given factor prices, industry demand, and basic 
technology because it governs the feasibility and effectiveness of vari- 
ous forms of rivalry. 

This explains why changing prices relative to those of rivals is not a 
frequent source of disturbance of balance. If the balance has existed for 
long, presumably the relative prices of sellers (including the different 
prices in different segments of the market by one seller) are such that 
market shares are not shifted radically at the initiation of either buyers 
or sellers. Presumably such a balance is optimal to the various firms 
under the existing conditions. 

Nevertheless, it is probably true that many firms could increase short- 
term profits by charging more, and perhaps even by charging less. The 
latter prospect is limited, however, not merely because of the expected 
reaction of rivals, but also by the fact that industrial and commercial 
buyers often give their regular suppliers an opportunity to match price 
cuts. That short-term opportunities to charge more are not seized re- 
flects the fact that firms are guided by their long-term revenue curves 
which are much flatter than are the short-term ones. Indeed, vis-a-vis 
rivals who occupy similar market positions, a firm may look upon that 
curve as though it were horizontal at least at outputs less than those 
prevailing. That it is perfectly consistent for a firm whose capital has 
been committed to think in terms of its short-term marginal cost curve 
and its long-term revenue curve has recently been pointed out by Har- 
rod. 4 

From the preceding argument emerges the conclusion that a change 
of price relative to that of the firm's rivals, that is, the disturbance of a 
balance by such a move, is appropriate only when the firm's market 
position is being changed correspondingly — a step which is not under- 
taken lightly. It is apt to succeed only when it is the result of a sub- 
stantial program, carried out over a period of time, which may involve 
a modification of almost every phase of the business, including its physi- 
cal plant. Consequently, changing prices relative to rivals tends to be 
part of a complex of almost continuous modification of product, pack- 
age, sales channel, guarantees, free services, etc., which characterizes not 
merely the markets for most finished consumer goods but also for in- 
dustrial equipment and many semimanufactured goods. 

R. F. Harrod, Economic Essays (Harcourt, Brace, 1952), p. 150. 


Less disturbing are forms of competition other than change of price 
relative to those of rivals, but the feasibility of these other competitive 
forms varies widely. Much advertising and other distributional effort is 
a means of standing still, not of moving forward. When used positively, 
devices other than price change do not often shift market shares rapidly. 
In part this is because their appeal to buyers is not so great and im- 
mediate as is a price cut. More typically, their lesser effect stems from 
the multiplicity and effectiveness of countering moves of rivals. Even 
trends in the strength of market positions, which are the result of com- 
petitive devices other than price, can be in operation for some time 
without setting off a price war. Consequently, there is nearly always 
some variation going on in the cost-price margins of particular firms or 
their market shares or the quality of their products, but not enough to 
disturb the fundamental balances of forces. 

It is in such a context as that just sketched that the stability of a 
balance must be appraised. In general, the less the product and its market 
conform to the specifications for a pure oligopoly (which really involves 
more relevant assumptions than fewness of sellers and standardization of 
product) the more likely it is that the important elements of the market 
positions of rivals will be such that they cannot be easily upset by feasi- 
ble competitive devices. It seems to me that most markets are of the 
latter sort, and that short of some development which warrants a stren- 
uous move by some participant, states of balance tend to be quite stable 
in the industrial sector of the economy. This does not mean a standstill, 
but rather an ability of industry to absorb many adjustments without 
open disruption. 


While orderliness, or tendency toward an equlibrium, is an important 
aspect of a market, the efficiency and the income distribution aspects of 
that equilibrium have properly been the center of attention in price 
theory. As applied here, this requires appraisal of the degree of opti- 
mality of the balance achieved among the forces of product quality, 
technology and cost, price, and output. In this section, we are not con- 
cerned with reactions to changes in general technology, consumer pref- 
erences and incomes, or factor prices, but, it must be repeated, the 
analysis is not abstracted from uncertainty with respect to these vari- 

Market Structure and Efficiency 

In light of the conditions for an adequate theory laid down earlier, the 
test for optimal industry performance cannot assume static conditions 
with respect to product, its costs, or its demand. We must be prepared 
to consider whether, with uncertainty present, the structure of the in- 
dustry inhibits or favors both the best product and the best method of 
production possible under the existing general state of technical knowl- 


edge. Or, conversely, the possibility must be considered that the struc- 
ture which has evolved reflects attempts to mitigate the impact of un- 
certainty of various possible origins. 

This brings us to the issue between two schools. On the one hand 
there is the more orthodox view that fewness of sellers and product 
differentiation lead to output at less than the optimum rate and deter 
progressiveness. The opposing view, most usually associated with Schum- 
peter and J. M. Clark, sees in this insulation the necessary conditions 
for the assumption of risks and uncertainties associated not only with 
innovation in product and process but also with other uncertainties 
related to investment in the most efficient equipment. 

It is not possible to resolve this basic issue on the basis of empirical 
findings. Objective appraisal is inhibited by the difficulties of delineating 
the various types of forces bearing on resource use in particular in- 
dustries. My own view — and it is only a judgment — is that there is a 
significant degree of truth in the idea that product quality and the struc- 
ture of costs will be socially more favorable in many industries if there 
is some insulation from overt price competition. But I do not think that 
this degree of insulation requires the current size distribution of sellers 
in some industries. 

Determinants of the Character of Co?npetition 

Beyond questions of the technology and costs are those of factors 
usually considered to determine the slope of the individual firm's reve- 
nue curve. Here the character of interfirm rivalry is explored as it is 
affected both by basic conditions of product demand and feasible tech- 
nology (to which all the firms in the group are subject) and by the 
discretionary actions of members of the industry. 

1 . Market Characteristics of the Product. The forms and effectiveness 
of competition are affected to an important degree by the market char- 
acteristics of the product. The categories developed in marketing litera- 
ture are particularly helpful, for they reflect not only the buying mo- 
tives and skill of final purchasers but also the role of the intermediate 
distributive trades with respect to various goods. These categories are: 
{a) "Specialty goods," which are those characterized by the fact that 
the buyer may first have to be convinced of the merits of the good be- 
fore he can be induced to purchase it and by the fact that, once con- 
vinced, the buyer will go to considerable difficulty to obtain the product 
of a particular producer. Excellent examples of specialty goods are the 
newer, "big-ticket" electrical appliances, (b) "Shopping goods," which 
are those bought after comparison of the quality of the products and 
prices of various suppliers. Excellent examples are piece goods, furni- 
ture, and fresh produce, (c) "Convenience goods," which are those pur- 
chased frequently and are often ordered by description or specification. 
Examples include both specialty items, such as toothpaste or prepared 
cereals, and standard items, such as sugar. 


The relative roles of the manufacturer and of the distributive trades 
vary widely among these categories of commodities partly because of 
corresponding variations in the competence of consumers as buyers. The 
manufacturer can with varying degrees of success build and hold final 
buyer preferences for specialty goods, but he is also dependent on the 
performance of his dealers. Retailers of the specialty subclass of con- 
venience goods tend to be reduced to the role of order-takers because 
consumers are not good judges of quality of these goods. On the other 
hand, the distributive trade tends to be in the saddle in the cases of 
shopping goods and standardized convenience goods. 

There is a strong tendency for the market characteristics of goods to 
change over time in such a way as to undermine any differentiation that 
yields abnormal profits. This means that specialty goods tend to become 
shopping goods or standardized convenience goods as they become part 
of the consumption pattern of most families and as the quality of the 
goods becomes satisfactory and quite uniform among suppliers. Like- 
wise, convenience goods of the specialty subtype tend to become fairly 
standardized as far as buyers are concerned. Few products are so stable 
in their physical attributes and in their use that they escape this process 

This evolution reflects to some degree growing buyer erudition but 
even that — and certainly the development of the satisfactory quality of 
the product itself — is a part of interflrm rivalry. In consumer goods the 
large retail organizations have had a major role in this dynamic and quite 
continuous process. 

Where this evolutionary process is not important, the stable market 
characteristics of the product, plus the degree of variability in what 
buyers want and the organization of the distributive trade, go far to de- 
termine the degree and character of the market segmentation which pre- 
vails. Segmentation does not get far for sugar; it is more important for 
canned goods; and it is a fundamental aspect of markets for the big-ticket 
products which combine specialty and shopping characteristics. 

2. Influence of Methods of Production and Structure of Costs. Here 
only certain aspects of the conditions of production and cost will be 
stressed, because the others are obvious. First, there is the question of 
the extent to which the method of production is a datum for the firm and 
the extent to which it may be altered upon the firm's initiative. One 
method may be more capitalistic than another and the related risks may 
be higher. Then, for the individual firm, the proportion of the total proc- 
essing, or of that plus raw material production and product distribution 
done by it, affects its cost structure. But there frequently are a number 
of feasible methods of vertical organization. Companies may engage in 
all or several vertical steps or they may merely assemble or blend parts 
or materials prepared by others with consequent effects on their capital 
requirements and on the relation of their fixed and variable costs. 

3. Is Market Structure Determinate? While the significance of au- 


tonomous determinants of the conditions of competition becomes more 
evident as one learns more about particular markets, their role must not 
be overstressed. The size distribution and other structural attributes re- 
flect the firm's past decisions. These may reflect mistakes, conscious 
moves or conditions long since gone, or conscious moves to control 
competition. Consequently, firms may be larger or more varied in their 
activities than the conditions of cost, of demand, and of uncertainty 
would require. Weighing the effect of such "artificial" structural ele- 
ments is difficult. 

My view is that the structure and operation of markets have a strong 
tendency to reflect the market characteristics of the product, the fea- 
sible conditions of its production, and the organization of adjacent indus- 
tries. Such factors influence even those industries that are made highly 
concentrated by design and dominate other industries still more com- 
pletely. But in no event must these conditions be viewed as fully de- 
terminate of either market structure or operations. Rather these forces 
define a zone within which management makes choices but of a sort 
which does not in most cases tie the hands of existing or potential rivals. 

Conditions and Effectiveness of Entry 

Four somewhat overlapping aspects of the entry problem should be 

1. Entry by established firms because of profit opportunities either 
in a field unrelated to the firm's initial activity or in an adjacent field or 
level is a widespread and a powerful competitive force. Movements into 
an adjacent industry may be either horizontal or vertical. Of these the 
vertical is the more promising, both because of the awareness of entry 
possibilities which a vertically adjacent firm is likely to have and be- 
cause such firms often possess the characteristic essential for successful 

2. Entry may be facilitated by the fact that most product lines and 
their associated markets are segmentized. Entry into some segments often 
is easy because they approach the purely competitive in character. Once 
such entry occurs, movement of that firm toward the segments where 
entry by a new firm would be very difficult may be, but not necessarily 
is, fairly easy. If it is, what might be an unscalable cliff if a new firm 
were to attempt to enter the difficult segment directly, will prove to be 
a set of stairs. 

3. Entry may be by the route of producing a substitute product or 
using better raw materials or manufacturing processes. Developments 
in chemistry, particularly, often can be used by the would-be entrants — 
frequently well-established firms — to enter with a product or process 
which can be substituted for those already in use. 

4. The effects of entry can be obtained at times by the assumption of 
a function usually carried out by firms engaged in another vertical step 


without actually owning the operation in that step. The most important 
case of this sort is the taking over by a distributive organization of the 
responsibility for distribution, including the branding, of a commodity 
which meets the buyers' specifications. 

Such potential sources and forms of entry have been widely employed 
in recent years, which leads to the question as to whether the economic 
effects of entry stem from the fear of entry or from the force of actual 
entry. The former has been stressed in the literature and has been sup- 
ported by some empirical evidence. In order for fear of entry to govern 
prices or to expedite process or product improvement, however, two 
conditions would have to exist. Potential entrants would have to be 
aware of the abnormal profits existing or of those obtainable by devel- 
oping a substitute product or channel of sale. Second, there would have 
to be quasi-agreement among established sellers, not merely as to the 
proper level of price, but also to refrain from themselves making such 
moves as introducing substitute products or channels of sale which would 
have the same effect as entry. 

Three observations, no one of which can be adequately documented, 
can be made relevant to whether management's fear of entry or the ac- 
tual occurrence of entry is the more important. For products which have 
the prospect of a long life, firms often hold prices at a level that will 
provide good but not high profits because of confidence of better long- 
run results. This policy reflects in part a desire to discourage entry, but 
not usually, I believe, in the sense of a specific appraisal having been 
made of entry prospects at different levels of price for the given product. 
Instead, a general policy as to margin over factory cost or estimated 
total cost, which may reflect experience as to margins which discourage 
entry, is applied to a wide category of goods. Second, specific potential 
entrants — usually companies or industries to which part of the product 
line is sold — may be "bought off," perhaps even by prices below total 
cost. Third, and most generally, it seems to me that entry actually oc- 
curs and forces down the price, or at least reduces sharply the share of 
the market that can be held by older firms without reducing price. The 
latter is important for consumer goods, for often entry is by a substitute 
product or by the creation of a separate market segment for a lower- 
priced but physically similar product. 

If there are vertically adjacent businesses which would be able, in 
terms of size, organization, and financial resources, to enter an industry 
and do not (or are not bought off by preferential pricing), this is prima 
facie evidence that this industry is performing competitively. Accord- 
ing to this test, meat packing would be judged to be competitive, but the 
test would not prove building materials manufacture to be so, because 
there is a lack of potential, vertically placed entrants into that business. 

What may be more feared than entry is the prospect that abnormal 
profits will lead firms already in the industry to expand shares by devices 


which fall short of overt price cuts and which are difficult to restrain or 
counter or, indeed, whose effectiveness is difficult to appraise ex ante but 
which nevertheless do force down realizations relative to cost. This is 
but part of the competitive conduct to which we now turn. 

Product-Quality and Cost-Price Effects of Nonprice Competition 

Maneuvers other than overt price change for an unchanged product 
have a major role in determining the economic attributes of equilibria in 
industrial markets. Let us assume a tacit (or even overt) agreement on 
a level of price which yields abnormal profits. Such an agreement does 
not of itself foreclose but rather encourages other moves such as prod- 
uct and process change, sales effort, and the development of new sales 

The list does not end there. In consumer goods industries, entry by 
some of the collaborators into lower-priced segments of the market 
tends to pull down prices there, and, as a result, the agreed-on prices in 
the more defensible and higher-priced segments become untenable. In 
other cases — particularly those of producers of standardized materials — 
integration into a fabrication or distribution stage facilitates not merely 
competition in the quality of the product and attendant services but also 
the making of price concessions. Just how rapidly and to what extent 
such forces work and how much they reduce realizations for the prod- 
uct line will vary widely. 

Tacit or even formal agreement about these methods of competition 
is improbable for the following reasons: 

1. In the use of these other variables, the output of various firms per 
unit of input does not necessarily tend toward uniformity. There is a 
substantial chance that with a given expenditure one firm may do much 
better with particular techniques of distribution and selling or be more 
successful in product research than will rivals. This contrasts with our 
usual assumption of a strong tendency toward equal output per unit of 

2. An important contribution to a firm's market position may result 
from being first or more active in nonprice competition, but rarely is 
that gain derived from being first in price competition. 

3. Alternatives to price competition involve maneuvers from which 
retreat is easy when desirable or from which a tangential move is feasible. 
Neither is true of a price cut. 

To the extent that these forms of competition are effective, noncom- 
petitive levels of prices and profits are subject to erosion. Much work 
has to be done before we know how rapidly this erosion occurs and how 
well it stacks up against potential alternatives. 

It must be evident, also, that the extent to which beneficial economic 
results stem from these forms of competition depends in large part on 


the degree to which a dynamic view of such markets is appropriate. If 
technology is an independent variable which would be fully utilized if 
far more overt price competition were engaged in, then the results of 
nonprice competition are not good. If, however, the degree of optimal 
use of technology is itself a reflection of the industry structure and the 
choice made of competitive methods, much good can be seen. Clearly, 
one's conclusions as to the validity of the Schumpeterian framework 
would vary among industries and, for a given industry, among stages 
in its development. 

There is also the second part of the Schumpeterian argument — which 
was postponed earlier — namely, the eroding of the insulated positions 
of established firms resulting both from the entry of large firms and 
from the nonprice competition provided by them. No comprehensive 
empirical study of this proposition has been made. My own opinion is 
that the erosion tendency is strong in a large portion of industrial situa- 
tions, that in some ways it is costly to the economy, but that these costs 
must be balanced against the feasibility of alternative routes to progress 
in an industrial society. 


The theoretical framework just sketched is really a generalized form 
of the long-standing analysis of the reaction of firms to profit oppor- 
tunities. Ordinarily, the emphasis has been on the case of the firm that 
has lower costs for a given product than its rivals and which, if rewards 
appear good compared to risks, would expand output. 

Here, however, the way in which even such a factor as entry can work 
must be adapted to the characteristics of industrial markets. Thus vari- 
ables held constant in a simple cost-price-output model must be intro- 
duced and their significance explored not merely by assuming them to 
be equivalent to a price change. When that is done, one finds that these 
variables have, potentially, a major effect on economic results and op- 
erate in spite of, and tend to undermine, a level of price for an unchanged 
product whenever that price fails to reflect an optimal combination of 
product quality, process and cost, and output rate. 

For a given product line, it is helpful to visualize a composite demand 
situation in which each item in the line has a demand curve of different 
elasticity. In addition, each item entails somewhat different direct costs 
of production and quite different general overhead costs. Different firms 
may be selling different items but usually each firm sells more than one. 
An optimum use of resources, in the sense of obtaining the full econo- 
mies of scale and the full use of capacity, is quite possible in such a case, 
even though the revenue curves in some segments are far from perfectly 
elastic. The reasons are, for static analysis, that the revenue curves in 
some segments are perfectly elastic and this tends to eliminate unused 


capacity. For dynamic analysis, the major point is that in an industry so 
structured, uncertainty may be sufficiently minimized that the optimum 
scale of plant will be actually established. 

Once one goes further and observes historical or dynamic processes, 
he will see that the relationships among segments need not be stationary. 
Revenue curves may, in fact ordinarily do, flatten out in the higher- 
priced segments, and changes in the product, in the process of its manu- 
facture, or in the channels of its sale, work to upset those price relation- 
ships among segments of the industry which do not reflect cost dif- 

When viewed this way the economic attributes of a balance cannot be 
appraised solely or even primarily by examining the cost-price-output 
relations for an unchanging, narrowly defined product. One must con- 
sider in addition the unmeasurable changes in product quality and the 
measurable effects of technological advance on the level of costs. A 
qualitative judgment must then be made as to whether the maximum 
economies of scale are obtained. Finally, the efficiency with which com- 
mitted resources are used and the division of the benefits thereof, that 
is, the price-cost-output relations which exist, must be judged for the 
entire family of items which go to make up the product line. Even then 
a single demonstration of cost-price-output relations cannot be made. A 
number of varied combinations and magnitudes of the variables stressed 
here are possible. Nevertheless, the general tenor of the argument, both 
as to the variables stressed and much of the empirical evidence referred 
to incidentally, is to suggest a strong tendency for economic results of 
industrial markets to be closer to the optimum than one would conclude 
from much of the literature about this sector of the economy. 


The general argument to be presented here is that factor price move- 
ments tend to be reflected in corresponding price changes but that de- 
mand changes do not. Obviously the degree of concomitance of move- 
ment of these two forces will modify the general statement just made. 
Both these modifications, if pursued, and the general statement just made 
find their explanation in the fact that firms' reactions to cost and price 
and demand developments reflect the structural position in which they 
find themselves and their relative market positions. 

Reaction to a Factor Price Change 

Factor price movements do not directly affect market positions of 
particular firms, and consequently selling price responses are not apt to 
be disruptive. In part this is because firms recognize the tendency to- 
ward uniformity in impact of these cost changes, although there may 
be some difference in timing of this impact due to variations in firms' 
degrees of vertical integration or in their inventory positions. More 


basically, this reaction is because market position has to do primarily 
with selling. Only indirectly, as some firms react to the factor price 
movements, are the conditions of selling market rivalry altered by a 
factor price movement. But when factor prices fall, the initiator of a 
selling price reduction is not suspected of trying to enlarge his share. 
Or, in reverse, the boldness of the firm which moves to reflect higher 
factor prices in his selling prices is appreciated, particularly when mar- 
gins have been squeezed sharply. 

Yet the differential immediate impact of factor price movements on 
firms and on market segments has much to do with the selling price re- 
sponse. Firms which are not integrated vertically or which are in a short- 
inventory position feel the factor-price movement more promptly than 
do rivals with the opposite attributes. In most product markets there are 
some segments — it happens that these also are the more nearly open mar- 
ket segments — in which margins between direct costs and selling price 
are much narrower than in other segments. Factor price movements im- 
pinge more heavily on such narrow-margin sellers and they are anxious 
to change their prices as soon as possible. 

When factor prices fall, the selling prices in the lower-margin seg- 
ments are apt to drop away from the quoted prices of the major-segment 
sellers and customers will be pulled away from the latter. This is the 
reason that major sellers are usually price followers on the downside. 

On the other hand, when factor prices rise these low-margin sellers 
feel the margin squeeze heavily. They edge up their prices as much as 
they can relative to those in the higher-priced segment. This reduces the 
substitution between the goods of the sellers in that segment and of 
those in the low-margin segment, a development which paves the way 
for the former to raise their prices overtly. When they do so they are, 
in a sense, again price followers rather than leaders. 

Another important influence of a factor price change is on the ex- 
pectations of intermediate buyers. When factor prices rise, a firm's cus- 
tomers need not fear a selling price decline. Hence, they are in a posi- 
tion to make an almost riskless speculation in inventories by stepping up 
their orders in anticipation of a price movement. If factor prices fall, 
business customers can hold off orders, a development which often tips 
the balance toward a selling price change. In that event, the problem 
associated with the kinked demand curve tends not to be present. In- 
deed, a factor price change usually does not disturb the internal balance 
within the group but merely affects the level of costs and prices about 
which day-to-day rivalry brings small but not disruptive changes in 
realizations, volume, and market shares of rival firms. 

All such conclusions must be tempered, however, according to the 
degree and frequency of factor price movements and the concurrent 
state of demand. The thesis just outlined obviously fits most closely the 
thin-margin industries whose buying prices vary frequently, such as 


those which process crude farm products. Their selling price changes 
are frequent and business customers have an almost perfect basis for 
speculative action. 

Where such experiences are infrequent or where the cost change is 
small relative to the feasible unit of price change, there is more hesi- 
tancy and even some tendency to use the indirect means of adaptation 
to be sketched later in connection with reaction to demand changes. 
Price increases are more difficult to bring about, and noncollusive price 
leadership or other means of implementing a meeting of minds are apt 
to arise. The same may hold where there has been an accumulation of 
cost changes for relatively unimportant items which add up to an im- 
portant reduction of profits. Of course, such small cost changes are often 
lost in a welter of product, channel-of-sale, and realization changes. 

Reaction to Demand Change 

Here attention focuses on a change of demand which, if followed by 
a roughly corresponding price change, would squeeze or enlarge the 
margin between direct costs and selling price. The marginal cost curve is 
assumed to be horizontal over a fairly wide range of volume rates. De- 
mand changes usually affect the relative market positions of firms be- 
cause the selling markets of firms are usually more varied in geographical 
coverage and in customer classes than are the earlier-stage markets for 
materials used in manufacturing. Some areas or some classes of customers 
will have experienced more or less variation in income or whatever is 
the cause of the change of demand. Consequently, the impact on sellers 
of the demand change tends not to be uniform and overt price changes 
would augment the effect. 

Both because a decline of demand creates excess capacity and hence 
each firm is unusually sensitive to loss of volume and because a price 
cut would then be viewed as a threat to market position, firms hesitate 
to initiate such a decline and react strongly to a price cut by a rival. In 
the first instance, firms tend to accept such volume and profits as the 
reduced industry demand will yield for the products, sales programs, and 
the price structure which were already in effect. Added intensity in 
carrying out the existing sales program may be tried. If the low demand 
is prolonged, selective price concessions may be made. Some firms may 
enter low-priced segments which they had avoided earlier. The degree 
to which such steps are taken is affected by the feasibility of these de- 
vices and by the proportion of costs which are fixed. Typically, overt 
price cuts which would squeeze the margin over direct costs will be 
made by the firms only when they become financially embarrassed or 
when the growth of price concessions and shift of volume to low-priced 
segments make the old level of prices in their major segments unten- 
able. Whether or not such price changes are made, the level of realiza- 
tion for the product line as a whole will fall relative to direct costs. 


Indeed, when demand falls, choices made by buyers among the items 
in a product line often reduce the level of realization without a change 
having occurred in the price for any one item. Particularly where con- 
sumer incomes have declined there is ample evidence that the propor- 
tion of long-margin items relative to short-margin ones falls. Conse- 
quently, the margin over direct costs for the product line as a whole will 
be reduced, and unless overhead outlays can be cut, profits will drop 

Except where the margin between direct cost and price has been 
squeezed by earlier developments, an increase in demand is followed 
by a process reverse to that just described. Here the idea of the kink in 
demand is particularly relevant. Short of sharp pressure from rising 
marginal costs (not higher factor prices), which may not occur until 
output exceeds that for which existing facilities were designed, no one 
seller would have reason to expect to be followed in a price increase. 
Aggregate profits of the firms are rising. Firms are inclined to use such 
an occasion to cement relations with customers, and failing to advance 
prices is a good method. Those whose market shares (and possibly mar- 
ket positions also) might appear to mark them as most likely to advance 
prices are often gaining by a shift within their line to the long-margin 
items and channels of sale. The firms who are lagging in these regards 
are least likely to follow a price increase. 

On what happens when sharply rising marginal costs are faced, our 
evidence is less conclusive. In more normal times than those of recent 
years — if there be such — demand increases are less sharp. On such oc- 
casions firms are apt to undertake a capacity expansion and to ration 
customers rather than to increase selling prices in the meantime. All such 
conduct is motivated by the desire to preserve and, if possible, to in- 
crease the strength of the firm's market position. 


In spite of modifications necessary for particular situations, most of 
which have not been spelled out here, the contrast between reaction to 
a change in demand and to a change in factor prices is valid and impor- 
tant. In direct proportion to the cost impact (and frequency) of change 
of factor prices, an approximately corresponding and overt change in sell- 
ing price can be expected. On the other hand, an overt change in price 
which would squeeze or expand the gross margin over direct costs for 
identical products does not ordinarily occur in response to modest and 
short-term movements of demand. Instead, other adaptations initiated by 
either sellers or buyers reduce or expand the general level of realiza- 
tion relative to cost for the varied items and channels of sale which make 
up the market for a product line. 

Indeed, a general rule, to which I think exceptions are few, is that the 
realization for the product line will always move in the direction of a 


change of demand whether or not associated with a movement of direct 
costs. The extent to which overt price moves, price concessions, changes 
in product and channel of sale, or variations in the proportion of volume 
made up of long- and short-margin items, respectively, will be the source 
of this up or down movement of the level of realization for the product 
line relative to cost, will vary widely. Any analysis which omits changes 
in this realization for reasons other than those noted or which omits trans- 
action-price changes even for narrowly defined products will lead to inac- 
curate conclusions. 5 


Definite and quantifiable conclusions cannot be laid out with respect 
to resource use in and welfare results of most industrial markets. Ob- 
viously these markets do not work like those whose structure approxi- 
mates that of the pure competition model. That is irrelevant, for the 
final interest is in results. It seems to me both evident and understandable, 
once one is versed in the whole of the structure and operations of mod- 
ern industrial markets, that their operating results approach more closely 
the ideal than one would surmise to be true from textbook models of 
oligopoly and monopolistic competition or from the literature on con- 
centration. This is the net result, I think, of sellers' reaction to uncer- 
tainty, of intersegment or even intermarket movement of firms or re- 
division of function, and of the wide opportunity for firms to gain 
individually by competitive devices other than overt price change — de- 
vices which after some lag tend to wipe out abnormally high margins for 
the product line as a whole. This means that I attach high value to the dy- 
namic view of most of these markets. The degree to which optimal 
aggregate results are obtained for these reasons (and also the effect of 
lesser uncertainty on the actual attainment of the economies of scale) 
must differ substantially among markets, and only the beginnings of the 
specification of the determinants of those degrees can now be made. Ob- 
viously, however, the market adjustments sketched here work differ- 
ently, often indirectly and more slowly, than one would visualize for an 
open market where there would be many buyers and sellers. 

Given that conclusion, one's appraisal of the performance of such 
markets depends in considerable degree on what he expects market 
processes to accomplish. If he thinks in terms of providing guides to 
individuals' acts, such as relative prices as a guide to choices by con- 
sumers, these markets fall far short of the ideal, as do markets which 
are imperfect for any reason. Furthermore, if one's system of economics 

5 The economic significance of the adjustment of the level of realization for a 
group of related items and a proposed method for measuring such realization changes 
have been considered by the present writer in "An Economic Appraisal of Price 
Measures," Journal of the A?nerican Statistical Association, December, 1951, pp. 


requires that prices respond directly and promptly to movements of de- 
mand, industrial markets stack up badly, for their adjustments in such 
cases tend to take place indirectly and laggardly. A reminder should be 
made, however, that these processes of indirect adjustment may bring 
cost-level and product-quality changes more rapidly than they would 
otherwise occur. Wherever these alternatives are feasible, they tend to 
bring about — slowly, to be sure — changes in the level of realization com- 
pared to cost for the product line as a whole and, to that degree, ef- 
fectuate the adaptation to changes of demand which is called for in 
economic theory. 

On the other hand, if one is concerned about a longer view of the 
performance of markets or concludes that only a longer view is ap- 
propriate for industrial markets, then his judgment of the performance 
sketched here is more optimistic. This latter view is what I take to have 
been the essence of the "workable competition" idea. In that sense, both 
the properties of equilibria and the reaction to factor price and demand 
movements in those industrial markets where effective, longer-run col- 
lusion does not exist, have a strong tendency toward a desirable use of 
resources and toward adaptability to major changes in conditions of cost 
and demand. 


Antitrust and the Classic Model* 


"Outsiders," wrote John Neville Keynes in introducing his Scope and 
Method, "are naturally suspicious of a science, in the treatment of which 
a new departure is so often and so loudly proclaimed." 1 It is a curious 
and disturbing fact that at the traditional center of economics — the role 
of markets in solving the general problem of order in economic life — 
new departures have been proclaimed so often. 

Outsiders become aware of economics mainly when economists ap- 
praise the working of the economy or propose guides for its control; 
and it is here that the theory of values and markets, more than in its 
abstract expression, has chief impact. In the quarter-century since Cham- 
berlin put monopoly and competition together in a single formulation, 
economists have stressed the prevalence of monopoly elements in mar- 
kets, but have differed in interpreting this condition, especially in its 
bearing on policy. Commonly they relate it to pre-Chamberlin thought 
regarding the nature and place of competition, and bring a putative 
"classic" model, at least implicitly, into their discussion. In doing so they 
will agree that such a model describes badly how markets really work, 
but will often disagree as to its normative role. They may accept its 
guidance in defining the necessary working of markets in a free system, 
and then despair of the future of control through markets that do not 
and cannot work in that manner. Or, more likely now, they may reject 
the model as a false guide because our present economy seems to do very 
well without meeting its requirements. As substitutes for it they may 
rely on theories of "workable competition" and "countervailing power." 

The present contention is that both of these views misrepresent earlier 
analysis when they deduce policy implications from models used in 
static theory. How wrong the deduction can be is apparent when we 
examine the work of such economists as John Bates Clark and Alfred 
Marshall who brought their theory into the practical realm. Their han- 
dling of market problems, running the gamut from static theory to policy 
proposals, warrants neither the pessimism nor the rejection. Indeed it 

* The American Economic Review, Vol. XLVII (March, 1957), pp. 60-78. Re- 
printed by courtesy of the publisher and the author. 
I University of Michigan. 

1 The Scope and Method of Political Economy (London, 1891 ), p. 8. 



would be more accurate to say that their thinking set a course which we 
are still largely following, and perhaps without being much farther down 
the road. The senior Keynes should find merit in a fuller sense of the 
continuity in this development. 



Galbraith, for example, proceeds in his American Capitalism from the 
position that a "vast distance separates oligopoly from the competition 
of the competitive model." "It is a measure of the magnitude of the 
disaster of the old system," he says, "that when oligopoly or crypto- 
monopoly is assumed it no longer follows that any of the old goals of 
social efficiency are realized." And again: "By evolution, from a system 
where nearly everything worked out for the best, economists found 
themselves with a system where nearly everything seemed to work out 
for the worst." 2 

As thus stated, this view of earlier thinking may lead to a denial 
merely of its descriptive validity, or also of its normative worth. In The 
Decline of Competition Arthur R. Burns displayed the first of these 
leanings. He began by adopting the usage common since Chamberlin of 
employing the term competition to mean pure competition and of re- 
ferring to situations departing therefrom, that is, to most markets, as 
noncompetitive or monopolistic. Under a variety of circumstances busi- 
ness firms act in light of "the effects of changes in their output, or their 
price policy, upon the market as a whole," and thus "find themselves in 
the position of a monopolist" in price and output decisions. In a con- 
text not of abstract theory but of market appraisal and policy recom- 
mendation, Burns finds it significant that "price and production policies 
would be expected to differ from those associated with perfect competi- 
tion." In his long and impressive recital of the monopoly elements in 
markets, it seems to be enough to show that these elements are present — 
unnecessary to determine how seriously they cause price and output 
adjustments to depart from an acceptable norm that embraces all ele- 
ments of public interest. More recently Burns has affirmed his view that, 
under the antitrust laws, "we have failed to achieve a competitive sys- 
tem at all closely resembling that which was in the minds of the econo- 
mists of the last century . . . ," and presumably has indicated his con- 
ception of the earlier view when he says: "Only pure price competition 
can produce the results which most people have in mind when they de- 
fend what they call in general terms 'the competitive system.' " 3 

2 J. K. Galbraith, American Capitalism — The Concept of Countervailing Power 
(Cambridge, Mass., 1952), pp. 45, 46, 51. See also chap. 2. 

3 A. R. Burns, The Decline of Competition (New York, 1936), pp. 3-6, 40-41. 
The second reference is to his contribution to the "The Effectiveness of the Federal 
Antitrust Laws: A Symposium," D. M. Keezer, ed., American Economic Review, 
Vol. XXXIX (June, 1949), pp. 691-94. 


But to Galbraith the lack of fit of the competitive model has quite a 
different meaning. While, according to it, things should work out for 
the worst under present conditions, they have not done so. The Ameri- 
can economy appears wonderfully effective. The model must thus be re- 
jected as guide as well as description; there can be no need of promoting 
the competition it envisages. In Galbraith's theory the offsetting power 
of groups on opposite sides of the market fills the breach, and market 
power is thereby kept from upsetting the allocative and distributive 

The equally optimistic and more widely accepted view of present 
markets is that competition, to succeed as regulator, need not approach 
the perfection of the model — indeed, it should not do so. This is the 
view that is now variously formulated under the caption of "workable 
competition." The essence of it is that even such rivalry as prevails when 
competitors are few can serve quite well in assigning resources and divid- 
ing income, and is greatly superior to its improbable purer cousin in 
promoting productivity and progress. In developing this view Schum- 
peter doubted the complete success of pure competition even as a static 
maximizing agent, but stressed mainly that an ideal disposition of re- 
sources at a given point in time is of small consequence when compared 
with the development of production through time. The competition 
that counts is "the competition from the new commodity, the new 
technology, the new source of supply, the new type of organiza- 
tion . . ."; and for it to operate, substantial elements of market power 
are necessary, not only as the concomitant of requisite firm size but as 
steadying influences in what would otherwise be too turbulent an eco- 
nomic sea. The present point is that Schumpeter regarded this theory of 
market operation as almost completely at odds with traditional thought. 
Despite some recognition of monopoly, "neither Marshall and Wicksell 
nor the classics," he said, "saw that perfect competition is the exception 
and that even if it were the rule there would be much less reason for 
congratulation than one might think." 4 

With similar optimism respecting big-firm capitalism A. A. Berle, in 
his recent The 20th Century Capitalist Revolution, joins in the rejection 
of older thinking, but with his own characteristic interpretation. Under 
the corporate system it is not true, he says, that "competition of great 
units (which does exist) produces the same results as those which used 
to flow from competition among thousands of small producers. . . . And 
it is indefensibly disingenuous to assert that these operations are prima- 
rily following economic laws more or less accurately outlined by the 
classic economists a century ago when the fact appears to be that they 
are following a slowly emerging pattern of sociological and political 
laws, relevant to the rather different community demands of our time." 

4 J. A. Schumpeter, Capitalism, Socialism, and Democracy (2d ed.; New York 
and London, 1947), pp. 74-78 and chaps, vii, viii. 


For urging the effectiveness of competition among the giants, Sumner 
Slichter is put in the category of the disingenuous, "since competition 
within the system of corporate concentrates produces results quite dif- 
ferent from the balanced economy expounded by Adam Smith." 5 What- 
ever the looseness of his history, whether of markets or of economic 
thought, Berle joins with Schumpeter and Galbraith in finding that 
present results are generally good. 

Galbraith and Berle wrote for a wide audience and Schumpeter has 
been widely read; and their rejection for present use of a supposed classic 
model is now echoed in abler segments of the public press. Thus Busi- 
ness Week, under the heading "Clobbering Theory," reports the 1953 
American Economic Association meetings in which Galbraith's thesis 
was discussed. "Classic economics teaches that only a competitive econ- 
omy can be sound and prosperous. . . . But the fact is that the United 
States economy bears only a remote likeness to the classic picture of a 
competitive system — and yet it has prospered enormously. . . ." 6 And 
Fortune, in an article on "The New Competition," points out that "the 
word competition no longer means what it once did." It is a competition 
that prevails even among oligopolists and it has been a "stunning suc- 
cess." This competition is said to be essentially different from what 
competition meant "to most of the economists and experts who have 
until recently shaped the accepted notions of competition. . . . Com- 
petition to them is a way of life that can be defined fairly rigidly. They 
conceive competition in terms of the grand old original or classic model 
of Adam Smith and his followers." 7 

To economists trained in the 1920's and before, as this writer was — 
and especially to economists who have long followed the theory under- 
lying antitrust policy — the foregoing oft-repeated view of what has hap- 
pened to economics must seem mildly shocking. Contrary to this view, 
it would be truer to say that the trend represented by the phrase "work- 
able competition" is a natural outgrowth of the thinking fifty years ago 
of such economists as J. B. Clark and Alfred Marshall. The policy con- 
clusions objected to by workable-competition advocates really rest on 
the broadened definition of monopoly in much more recent theory. 
Some would recast earlier theory in line with this definition, but Mar- 
shall and Clark, and other theorists of realistic outlook from Adam 
Smith on, quite surely would have rejected the policy implications of 
any such reconstruction of their thought. Their conception of welfare 
was never confined to the norm of precise efficiency in allocation and 
their practical judgments moved well beyond the narrow boundaries of 
static analysis. 

5 A. A. Berle, Jr., The 20th Century Capitalist Revolution (New York, 1954), 
pp. 11-12,43-52. 

6 January 9, 1954, pp. 93-99. 

7 June, 1952, p. 99. 



When economists give their main attention to a theoretical problem 
they are not in effect declaring that the solution of that problem meets 
all the requirements, or even the principal requirements, of well-being. 
Economists of the nineteenth century, especially the classical, Austrian, 
and neoclassical economists, gave their greatest effort to the problem of 
value. Smith, Ricardo, and the classical writers were unable to relate 
utility and cost meaningfully, or the values of factors and products 
fundamentally, and it required generations of economists to define the 
value conditions of economic behavior. The problem was fascinating in 
itself, and it seemed vital because prices were an obvious basis of action 
and of income in a specialized and exchanging society. A full grasp of 
the system-wide problem of economy in using resources, and the in- 
herent relation of distribution to it, seems to have come only gradually, 
and with it a full realization of the role of a system of values in solving 
the general problem of order; but as thinking acquired this focus, the 
necessity of dealing with so complex a problem under simplified assump- 
tions became evident. Very naturally the analysis was carried forward, 
especially by mathematically inclined economists, to an attempted final 
formulation of the condition of ideal maximization in the whole econ- 
omy. The problem was worthy of the effort given it; but the inference 
is unwarranted that economics thereby limited its concern to a nice 
allocation of resources, or viewed the assumptions of allocation theory 
as descriptive of the real world. 

The supposed need today of a new theory of economic performance, 
at sharp variance from the traditional, may indeed reflect some shift in 
emphasis among the several economic goals. There is great stress now 
on progress in total output, and earlier theory may have dealt inade- 
quately with it. Schumpeter doubtless judged rightly in asserting the 
superiority of this objective over the niceties of assigning and rewarding 
factors; but he was wrong, nevertheless, in imputing neglect of it to the 
leading economists of earlier periods, or the advocacy by them of con- 
ditions which would impede its accomplishment. These economists 
seemed to think that coordination of activities through markets was 
the most fruitful problem for economists to attack; but most of them 
took it for granted that welfare depends primarily on high output and 
on the conditions necessary to it. Adam Smith struck the keynote in 
beginning his Inquiry into the Nature and Causes of the Wealth of Na- 
tions with a treatment "Of the Causes of Improvement in the Productive 
Power of Labour . . ." and in his obviously greater concern over "Prog- 
ress of Opulence" than over any model that might now be described as 
"grand old original or classic." J. S. Mill concluded the introduction to 
his Principles with the statement: "The laws of Production and Distri- 


bution, and some of the practical consequences deducible from them, 
are the subject of the following treatise." His long first Book deals with 
production; and when, following Books II and III on distribution and 
value, he passes in Book IV from the "statics of the subject" to its 
"dynamics," he views the progress of society largely in terms of expand- 
ing production. 8 

No one should attempt to state briefly the criteria by which A4arshall 
was guided in the incidental appraisals that appear in his broad picture 
of economic organization. Advance in output is implicit as a leading ele- 
ment in many of his remarks on progress; but his main concern was 
with nothing less than the whole improvement of man. Certainly it would 
misrepresent him grossly to say that he thought the success of a system 
depended on achieving certain marginal relationships in using resources. 
"The main concern of economics," he said, "is thus with human beings 
who are impelled, for good and evil, to change and progress. Fragmen- 
tary statical hypotheses are used as temporary auxiliaries to dynamical — 
or rather biological — conceptions: but the central idea of economics, 
even when its Foundations alone are under discussion, must be that of 
living force and movement." 9 

In the present context John Bates Clark is the most instructive spokes- 
man for traditional economics, since he was both an eminent expositor 
of neoclassical doctrine and a leading student of monopoly and antitrust 
policy, writing in the midst of early excitement over the threat of con- 
centrated market power to the economic structure. In formulating in 
his Distribution of Wealth his static, perfect-market theory of factor 
pricing and resource use, he characterized his effort with the concluding 
statement that "all real knowledge of the laws of movement depends 
upon an adequate knowledge of the laws of rest." He saw this approach 
as only a part of economics since "a static state ... is imaginary. All 
natural societies are dynamic; and those which we have principally to 
study are highly so." "A theory of disturbance and variation," he said 
then, would be "included in the science of economic dynamics; but the 
most important thing that is included in it is a theory of progress." 10 
Clark essayed this larger task in his Essentials of Economic Theory, 
though only as a "provisional statement of the more general laws of 
progress"; and in it he set forth his views of monopoly and of related 
policy, as will be noted below. 11 For the moment we need only observe 

8 J. S. Mill, Principles of Political Economy (New York, 1883, from 5th Lon- 
don edition), Vol. I, p. 42; Vol. II, pp. 271-72. As to his personal choice among the 
tests of a good system, Mill said he could not "regard the stationary state of capital 
and wealth with the same unaffected aversion so generally manifested toward it by 
political economists of the old school" (Vol. II, p. 336). 

9 Alfred Marshall, Principles of Economics (8th ed.; London, 1920), p. xv. 

10 The Distribution of Wealth (New York, 1899), pp. 442, 31, 33. 

11 Essentials of Economic Theory — as Applied to Modern Problems of Industry 
and Public Policy (New York, 1907), p. v. 


Clark's perspective of goals, as when he stated in his The Control of 
Trusts that ". . . progress is in itself the summum bonum in economics, 
and that society is essentially the best which improves the fastest." 12 

But it is only part of the story to see that leading economists of the 
past gave no special pre-eminence to the value matters they studied so 
thoroughly, and looked on their static hypotheses as something less than 
realistic description. While they stressed high and growing production, 
did this result not depend in their theory on a degree of competition 
impossible under modern conditions? Galbraith, notably, rests his criti- 
cism on this interpretation. Among "the old goals of social efficiency" 
he includes "getting the most for the least — the common engineering 
view of efficiency," together with "appropriate incentive to change — 
the adoption of new and more efficient methods of production"; and he 
lumps these production goals with those of optimum allocation and dis- 
tribution in declaring that when competitors are few "it no longer fol- 
lows that any of the old goals of social efficiency are realized." Indeed, 
for all of these goals to be reached, as Galbraith interprets the require- 
ments of earlier theory, competition should be construed even more 
rigorously than was done; and he applauds the more recent economists 
who "began to require of competition a meaning which would cause it, 
in turn, to produce the economic and social consequences which earlier 
economists had associated with it." 13 

The opposite view seems more plausible. Only when traditional eco- 
nomics is thus "perfected" is it vulnerable to the charge that, by its 
rationale, everything should work out for the worst in modern capital- 
ism. We should not inflict on it so damaging a refinement. It is true that 
competition has always been assigned a central place among the condi- 
tions of productivity and progress; it has been counted on to spur im- 
provement, rid industries of weak producers, prevent gain through re- 
stricting performance. But the competition that serves these ends need 
not be perfect — in major respects it should not be — and earlier econo- 
mists knew this, much as we do. Whatever the perfection formerly 
thought necessary for markets to perform certain allocative and dis- 
tributive functions — the topic of the following section — no such condi- 
tion should be read into their analysis of the more dynamic, onmoving 
aspects of economic performance. 

Again, J. B. Clark's thinking is pertinent — a natural source for stu- 
dents who would link their thinking with the past. Clark did indeed in- 
sist that monopoly is decidedly "unfavorable to continued improvement 

12 The Control of Trusts (New York, 1901), pp. 82-83. 

13 Galbraith, op. cit., pp. 16-20. Very properly Galbraith includes among the main 
economic goals— along with high productivity, effective allocation, and acceptable 
distribution — the stable high-level employment of resources. In dealing with this 
goal traditional economics comes off less well; but here the matter at issue is only 
incidentally, if at all, the competition that was assumed. 


in the productive arts" whereas "competition is the assured guarantee 
of all such progress." But Clark wrote prior to that unfortunate usage 
by which all that is not pure competition is labeled monopoly. By mo- 
nopoly he meant unified control of a market, and by competition, in 
this context, "healthful rivalry in serving the public." He feared the 
trusts that were developing and saw that "partial monopolies" were prev- 
alent and dangerous. 14 But he saw the advantages of large establishments 
and also of consolidations, including even their contributions to research 
and innovation, which Galbraith says was slow to be recognized. The 
following may summarize Clark's view: 

A vast corporation that is not a true monopoly may be eminently progres- 
sive. If it still has to fear rivals, actual or potential, it is under the same kind of 
pressure that acts upon the independent producer — pressure to economize la- 
bor. It may be able to make even greater progress than a smaller corporation 
could make, for it may be able to hire ingenious men to devise new appliances, 
and it may be able to test them without greatly trenching on its income by 
such experiments. When it gets a successful machine, it may introduce it at 
once into many mills. Consolidation without monopoly is favorable to prog- 
ress. 15 

Thus, in the manner of Schumpeter and others, Clark was saying not 
merely that productivity and progress can persist in the face of an ad- 
mixture of monopoly, but that within limits they are promoted positively 
by it. This view appeared most clearly in his appreciation of patent 
policy: "If an invention became public property the moment that it was 
made, there would be small profit accruing to any one from the use of 
it and smaller ones from making it. . . . This fact affords a justification 
for one variety of monopoly. . . . Patents are a legal device for pro- 
moting improvements, and they accomplish this by invoking the princi- 
ple of monopoly which in itself is hostile to improvement." He recog- 
nized the possibility of abuses, but he sensed the principle, which he 
stated elsewhere more abstractly, that perfect competition instead of 
being a condition of progress would actually prevent it. 16 

There is another quite different respect in which the fullest competi- 
tion is often deemed harmful and unworkable. It lies in the fact that in 
modern industry with its indivisibilities, fixed costs, and lumpy expan- 
sion in anticipation of demand, wholly unrestricted competition is likely 
to make profit seeking too difficult, losses too prevalent, for firms to 
remain healthy and vigorous. This view was also common a half-century 
and more ago and was urged as a reason for accepting, though with 

14 Essentials, pp. 364, 374, 382, 533-34. 

15 Ibid., p. 534. 

ie lbid., pp. 360, 366, 373. Clark's "five organic changes," the basis of his "economic 
dynamics," included growth of population and of capital and changes in methods 
and organization of production and in consumers' wants. Shifting of production to 
new products did not receive separate recognition but appeared under the last 
heading. Ibid., 203-6. 


misgiving, the limitations that size and combination bring. In his early 
work, The Philosophy of Wealth, Clark applauded the "conservative 
competition in which economists of a few years ago were able to see 
realized a general harmony of social interests"; and with it he contrasted 
"the fiercer contest in which eventual success comes to a participant 
through the extermination of rivals, the process well-named 'cut-throat' 
competition." "Easy and tolerant competition," Clark said, "is the an- 
tithesis of monopoly; the cut-throat process is the father of it." 17 

Later he provided an explanation, as we would now, in terms of fixed 
costs and unused capacity and the possibility that competition will drive 
prices down close to the level of variable costs. Such competition he 
pointed out, in discussing water transportation in the Essentials, results 
usually in "a merely tacit agreement to 'live and let live' "; and he thought 
"a normal kind of competition will stop short of the warfare which 
drives both rivals into bankruptcy." 18 Still later, in the second edition of 
The Control of Trusts, the situation in industry in general was ex- 
plained, and the case presented and conditions set forth for "a tolerant 
and normal competition" under which big industry can remain vig- 
orous. 19 This fear of wholly unrestricted competition was quite general 
among American economists of the period. 20 That Marshall questioned 
the wisdom of unlimited competition was evident both in his Principles 
and in his Industry and Trade, as will appear incidentally in the follow- 
ing section. 


Now let us turn, in this comparison of past with present thinking, 
from the conditions of expanding productivity and general industrial 
health to the distortions commonly attributed to monopoly. To say that 
a system with some mixture of monopoly in its competition may do a 
good job in developing productive power is not to say that it escapes 
serious misallocation and exploitation. Much that is now claimed on be- 
half of a new concept of competition, supposedly different from that of 
the older economists, amounts to saying that our economy does so well 
in expanding output that we can afford to overlook the distortions. This, 
in part, is what Schumpeter said. But theories of "workable competi- 
tion," as of "countervailing power," go further and give reasons why 
departures from purity in competition do little harm to price relation- 

17 The Philosophy of Wealth (Boston, 1886), p. 120. 

18 Essentials, pp. 414-15. 

19 J. B. Clark and J. M. Clark, The Control of Trusts (rev. ed.; New York, 1912), 
pp. 168-83. 

20 See, for example, A. T. Hadley, Economics (New York and London, 1896), 
chap. 6; J. W. Jenks, The Trust Problem (rev. ed.; New York, 1905), pp. 140, 
16-20; F. W. Taussig, Principles of Economics (2d ed.; New York, 1911), Vol. II, 
pp. 434-36. 


ships. Are these reasons much different from the views of economists of 
fifty or more years ago? Again, the present theme is that they are not — 
that they differ only as more fully developed concepts differ from their 

To begin with, we need perspective of the place that pure competition, 
or whatever the "classic model" implies, had in relevant earlier thinking. 
A plausible conclusion, when we scan the long attack on value problems, 
is that particular features and degrees of competition had a much smaller 
place in the whole analysis, even implicitly, than is now often supposed. 
The main struggle of economists over a century or more was not in 
spelling out marginal refinements but in putting the main building blocks 
in some sort of order: in relating utility and cost; in recognizing other 
costs than labor; in seeing the broad dependence of factor prices on 
product prices, as well as the narrower reverse relationship; in assigning 
separate values to factors used in combination — all, of course, with in- 
cremental logic but with the main structure transcending the static ni- 
ceties. In a society faced with vast possibilities of gross error in adapting 
complex resources to satisfying countless wants, formulation in value 
terms of the main elements of system-wide order was the goal to be 
sought and is the achievement now to be applauded. 

In this setting the problems of allocation and distribution do not de- 
pend for their solution on a certain kind of competition; the essential 
solution is largely independent of the type of market. It is often said 
that competition is the regulator of a market economy; but, on the con- 
trary, the chaos that would prevail in the absence of effective control is 
obviated not by competition alone but by the more general operation of 
the whole price mechanism. This is evident when we observe that even 
a monopolist can derive revenue only by producing what people will 
buy, and that he is best off when he aims at the most valuable flow of 
products from the resources he uses. Nor can he get the most profit 
without employing effective techniques of production, and in other re- 
spects selecting and combining factors to best advantage. His demand 
enters into the total demand for each factor, and this demand, with the 
supply, sets the price of the factor and the cost of its use, and thus pro- 
vides the essential barrier to inferior applications of it. True, his market 
power creates stresses that prevent full maximization from the social 
standpoint and his income may be greater than his performance requires. 
But the main elements of order are still present. A degree of competition 
that will keep these distortions within acceptable bounds has always 
been thought of as an essential part of the mechanism, but it is only a 
subordinate part of the whole scheme of control that the older eco- 
nomics explained for us. 

This is elementary and is said only because lack of perspective re- 
garding it has been common and has appeared in widely influential writ- 
ing. The confusion may be explicit, as in assuming that a market econ- 


omy is practically unregulated unless competition approaches perfection 
— or that the older economics held this to be the case. Or it may be 
implicit, as, for example, in Berle's contention that corporate operations 
are no longer guided by the market forces of traditional theory but by 
the mores of essentially political entities. Berle is offering a plausible 
theory of behavior within a limited range of decision making; but, so far 
as traditional economics is concerned, he overlooks the main point, 
namely that the greatest corporation is still subject to the tyranny of 
the income statement and can prosper only as it directs production in 
keeping with buyer preferences and uses resources with an eye to costs 
determined in a setting of alternative uses. 

Thus viewed, this value structure defines the broad conditions of 
order in a market economy even when competitors are few. But its 
formulators may still be charged with setting up misleading guideposts 
if, implicitly, they made the ultimate niceties the crux of the allocation 
process. This, however, does not seem to have been the case, at least 
among leading economists who related theory to practical issues. 

In this connection also, J. B. Clark is the best example of an able 
theorist concerned specifically with monopoly and antitrust. Clark seems 
generally to have recognized the difficulties that economists of the pe- 
riod are now said to have ignored, and his resolution of them suggests 
much present thinking. In his Essentials he says: 

The most striking phenomenon of our time is the consolidation of inde- 
pendent establishments by the forming of what are usually called trusts; and 
this and all the approaches to it are precluded by the static hypothesis. There 
is a question whether, after competition has reduced establishments in one sub- 
group to a half dozen or less, they would not, even without forming a trust, 
act as a quasi-monopoly. 21 

He saw the danger: "What we have is neither the complete monopoly 
nor the merely formal one, but one that has power enough to work in- 
jury and to be a menace to industry and politics." But markets still pro- 
vide protection: even when the entire product comes from a single 

. . . the price may conceivably be a normal one. It may stand not much 
above the cost of production to the monopoly itself. If it does so, it is because 
a higher price would invite competition. The great company prefers to sell all 
the goods that are required at a moderate price rather than to invite rivals into 
its territory. This is monopoly in form but not in fact, for it is shorn of its 
injurious power; and the thing that holds it firmly in check is potential compe- 
tition. . . . Since the first trusts were formed the efficiency of potential compe- 
tition has been so constantly displayed that there is no danger that this regula- 
tor of prices will ever be disregarded. 22 

But, said Clark, this "check works imperfectly. At some points it re- 
strains the corporations quite closely and gives an approach to the ideal 

21 Essentials, p. 201. 

22 Ibid., pp. 380-82; italics in original. 


results, in which the consolidation is very productive but not at all op- 
pressive; while elsewhere the check has very little power, oppression 
prevails, and if anything holds the exactions of the corporation within 
bounds, it is a respect for the ultimate power of the government and an 
inkling of what the people may do if they are provoked to drastic ac- 
tion." He was hopeful that a policy aimed at "keeping the field open for 
competitors" might obviate more drastic action. This would require pre- 
vention of unfair and predatory methods. "The preservation of a normal 
system of industry and a normal division of its products requires the 
suppression of all those practices of great corporations on which their 
monopolistic power depends." 23 

While Clark saw the possibility of quasi monopoly when competitors 
are few, he believed "that competition usually would, in fact, survive 
and be extremely effective among as few as five or six competitors, till 
they formed some sort of union with each other." 24 No well-formed 
theory of oligopoly governed his thinking, and thus he saved himself 
the ordeal, first, of assuming full joint profit maximization when there 
are few competitors, and then of finding later a complex of reasons for 
doubting whether in fact this outcome is likely under dynamic condi- 
tions. With only moderate skepticism he might have accepted Schum- 
peter's view that the monopoly elements may be just sufficient to offset 
the forces of "creative destruction" which threaten the disappearance of 
profit. Clark observed: 

The actual shape of society at any one time is not the static model of that 
time; but it tends to conform to it, and in a very dynamic society is more 
nearly like it than it would be in one in which the forces of change are less 
active. With all the transforming influences to which American industrial so- 
ciety is subject, it today conforms more closely to a normal form than do the 
more conservative societies of Asia. 25 

Marshall, like Clark, fits poorly the supposed pattern of older thinking 
that we are questioning — a pattern of implied optimism respecting cap- 
italism mistakenly resting on its assumed close resemblance to some 
model of near-perfect competition. Marshall was at least in part aware 
of the theoretical import of that ideal competitive state in which pro- 
ducers sell "in a large open market in such small quantities, that cur- 
rent prices will not be appreciably affected by anything which they may 
do or abstain from doing . . ."; 26 but he disliked pushing his hypotheses 
so far. For better or worse, a "principle of continuity," as he had called 
it, animated his thinking, and he saw "great mischief" in "drawing broad 

23 Ibid., pp. 383, 395. 

24 Ibid., pp. 201-02. 

25 Ibid., p. 197. 

26 Alfred Marshall, Industry and Trade (London, 1927), p. 401. Much of this 
volume was written long before its appearance in 1919, part of the type having been 
set in 1904. 


artificial lines of division where Nature has made none." 27 Thus, even 
in his Principles, in explaining "normal" pricing in manufacturing and 
merchandising, he did not adopt as his starting point the fluid, market- 
determined pricing of agriculture, but instead assumed the condition of 
quoted prices which prevails in such markets, with each seller dependent 
not on an impersonal body of purchasers but on specific patronage. For 
instance, in considering the common situation of firms that immediately 
must operate below capacity, he says: 

In a trade which uses very expensive plant, the prime cost of goods is but a 
small part of their total cost; and an order at much less than their normal price 
may leave a large surplus above their prime cost. But if producers accept such 
orders in their anxiety to prevent their plant from being idle, they glut the 
market and tend to prevent prices from reviving. In fact however they seldom 
pursue this policy constantly and without moderation. If they did, they might 
ruin many of those in the trade, themselves perhaps among the number; and in 
that case a revival of demand would raise violently the prices of the goods pro- 
duced by the trade. Extreme variations of this kind are in the long run bene- 
ficial neither to producers nor to consumers; and general opinion is not alto- 
gether hostile to that code of trade morality which condemns the action of 
anyone who "spoils the market" by being too ready to accept a price that does 
little more than cover the prime cost of his goods, and allows but little on ac- 
count of his general expenses. 28 

Thus, in his theory of short-run use of plant capacity, Marshall im- 
plicitly rejects pure competition as his expository framework, and he 
rejects also the allocative result of pure competition as a sufficient cri- 
terion in judging control through markets. 

Marshall's Principles was devoted to what he called "foundations," 
the exposition of the "normal" in equilibrium terms; and the more "bio- 
logical" approach required by the development of modern industry he 
put aside for separate analysis in his Industry and Trade?® In this analysis 
the prevailing theme is that even "open" markets display only a qualified 
competition, and monopoly and competition "shade into each other by 
imperceptible degrees." "Every manufacturer, or other businessman, has 
a plant, an organization, and a business connection, which put him in a 
position of advantage for his special work"; and thus "for the time being 
he and other owners of factories of his class are in possession of a partial 
monopoly. . . . Combinations for regulating prices aim at consolidating 
provisionally this partial monopoly, and at putting it in good working- 
order. . . ." 30 In this setting he examines at length the growth of plant 
size for technical reasons and the many-rooted development of corporate 
combination and cartelization in Germany, Britain, and America. 

27 Principles, preface to the 1st edition as appearing in the 8th, p. ix. 
28 Ibid., p. 375. 

29 As explained in the preface to the 8th edition of the Principles, pp. xii-xiv. 

30 Industry and Trade, pp. 178, 196. Marshall's extensive pre-Chamberlin exposi- 
tion of the theme of monopolistic competition is set forth by H. H. Liebhafsky in 
"A Curious Case of Neglect: Marshall's Industry and Trade," Canadian Journal of 
Economics and Political Science, Vol. XXI (August, 1955), pp. 339-53. 


Competition has a central place in neoclassical theory; but Marshall, 
its great exponent, remained unexcited by his impressive evidence that 
competition is manifest mainly in rough approximate ways. Monopoly 
power is a continual threat, but "absolute monopolies," he believed, "are 
of little importance in modern business as compared with those which 
are 'conditional' or 'provisional' " and the latter keep their place only if 
"they do not put prices much above the levels necessary to cover their 
outlays with normal profits." Like Clark, A4arshall thought a "severely 
monopolistic price policy" unlikely because "a man of sound judgment 
. . . will keep a watchful eye on sources of possible competition, direct 
and indirect." Potential competition, the competition of substitutes, a 
long-run concern over the welfare of customers, were all stressed as 
significant restraints. 31 

Marshall saw also the restraining effect of product differentiation. 
Though it violates the purity of competition, it obstructs all arrange- 
ments for price control, which are difficult to bring about when products 
are not standardized. At the same time Marshall observed that stand- 
ardized goods, which include "raw materials or half-finished products, 
or implements" used in business, are likely to be bought by firms that 
possess market power and "therefore are likely to meet the danger of 
oppressive action on the part of a combination, in control of things 
which they need to buy, by a counter-federation of their own. That is 
apt in its turn to stimulate the growth of similar federations on the part 
of traders or producers who need to buy some of their products; and so 
on till the end of the chain. . . ," 32 So "countervailing power," and the 
condition tending to produce it, received a respectful nod. 

Nothing like the "classic model" seems to have been considered seri- 
ously by Marshall as a policy goal. Early industrial competition, back to 
Ricardo, had not resembled any sort of market ideal, though the con- 
trary is now often assumed. It was rather the "aggressive competition" 
of "crude, though energetic men," a "species of warfare," and was not 
likely, as the preceding section indicated, to produce a "solid prosperity." 
For most of British industry Marshall found adequate the more re- 
strained kind of competition, with the greater admixture of monopoly, 
which came in his day. Even in America, he thought, "Anglo-Saxon 
moderation and stability have enabled competitive and monopolistic 
abuses to be kept within relatively narrow limits, with but little direct 
intervention of authority." At the same time, monopoly was more dang- 
erous than was generally realized, with greater menace in "monopo- 
listic association" than in "monopolistic aggregations"; and a policy more 
positive than publicity, which he generally favored, might become nec- 
essary. 33 

With such spokesmen as Clark and Marshall writing in this vein, it 

31 Industry and Trade, pp. 395-98, 405-09, 523-26. 

32 Ibid., p. 549. 

33 Ibid., pp. 179, 656, 400. 


is surprising that Schumpeter should have belittled neoclassical doctrine 
as he did. To him the competition of his predecessors was a "competition 
within a rigid pattern of invariant conditions," and not at all the com- 
petition of new and better products and processes that he thought im- 
portant. He failed to note that this emphasis of his was essentially an 
unfolding of earlier thought and that he was quite in the earlier vein in 
saying of this latter competition that it "acts not only when in being 
but also when it is merely an ever-present threat . . ." and that "in 
many cases, though not in all, this will in the long-run enforce behavior 
very similar to the perfectly competitive pattern." 34 It is surprising, too, 
that Schumpeter, in these allusions to traditional economics, failed to 
credit it with explaining the broad market and value structure that his 
theory implicitly relied on in circumscribing the distortions of crudely 
competitive markets. He saw the older economics not in its whole rele- 
vant expanse but only in its effort to sharpen particular relationships 
with the tools of static theory. In another context, however, that of his 
"socialist blueprint," he paid neoclassical economics the ultimate tribute 
of relying almost step by step on its essential structure in showing how 
socialism may solve the general problem of economy in using resources. 35 


There has been looseness at all times in perceiving the role of static 
models of competitive market operation. Such models are useful, indeed 
essential, in rendering manageable the numerous elements in the general 
problem of order in the economy. They supply the framework for trac- 
ing allocative effects of given practices and policies, and in this role 
provide a starting point in observing when market power is manifest. 
But static models may also mislead: through being supposed to reflect 
closely the actual processes of markets; through suggesting that they 
embrace all elements of welfare and afford a basis for judging economic 
performance as a whole; through tempting the user to toss all departures 
from their exact conditions into a common pot called monopoly and 
leading him, without guidance as to the seriousness of the deviations, into 
unhappy conclusions as to how the economy is working, or should be 
expected to work. 

However, economists who seem at times to insist in supposed tradi- 
tional fashion on near-perfect competition as a condition of acceptable 
economic performance may not carry this insistence into their more 
practical judgments. Galbraith supports his conception of earlier thinking 
by repeated use of Hayek's statement in The Road to Serfdoin that "the 
price system will fulfill this function [of general control] only if com- 
petition prevails, that is, if the individual producer has to adapt himself 

34 Schumpeter, op. cit., pp. 84-85. 

35 Ibid., chap. 16. 


to price changes and cannot control them." But the context of this state- 
ment does not imply purity of competition, since Hayek is only declar- 
ing the general superiority of control through markets over "central 
planning for the growth of our industrial system," which, he says, is by 
comparison "incredibly clumsy, primitive, and limited in scope." And he 
indicates that he finds acceptable a competition that can ordinarily be 
reconciled with the economies of size — one in which the firm, while it 
cannot control prices, can certainly influence them. 36 Pigou, with his 
elaborate concern over deviations from equality in marginal social net 
products, may likewise be thought of as intolerant of imperfect market 
adjustments. But Pigou was explicit that "simple competition," as he 
called it, is not feasible technically; and he preferred a limited antitrust 
approach, such as Clark favored, to a more drastic control of business. 37 

But even though they were not purists in their conception of adequate 
competition in a policy context, should not all these exponents of older 
thinking have been overwhelmed by the full impact of modern oligopoly 
theory? While granting some latitude to business decision in a dynamic 
system, could they digest the idea that business firms, separate but few 
in number, may so calculate each other's moves that they arrive at the 
price and output conclusions of the single monopolist? However loosely 
their frame of thought is construed, can modern markets be made to 
operate successfully within it? 

The answer is yes, if we accept Clark and Marshall as spokesmen of 
earlier thinking and are not ourselves overwhelmed by the first approxi- 
mations of modern theory. Indeed, without certain present insights, 
they came close to the spot where we now find ourselves, as conflicting 
insights begin to cancel out. Unworried by the neat logic of joint profit 
maximization when competitors are few, they were not bound to in- 
vestigate the exacting and unusual conditions of that logic: its assump- 
tions of a common view of demand and cost functions, of lack of aggres- 
sive desire for a larger share of markets, of standardization not only of 
products but of market terms in general, of pricing that is open and 
above-board, of absence of fear of new entries and substitute products 
and all the dynamic hurly-burly that Schumpeter made the center of 
his thinking. About the same position can be reached, in an unsophisti- 
cated way, without first falling into the oligopoly trap and then freeing 
one's self from it. 

Thus the views mainly to be corrected are not those of the older 
economists. They had a fair sense of the impact of modern industry, 

36 F. A. Hayek, The Road to Serfdom (Chicago, 1944), chap. 4. The first quota- 
tion is from p. 49 and is used by Galbraith, op. cit., pp. 15, 35. 

37 A. C. Pigou, The Economics of Welfare (4th ed.; London, 1950), chap. 21. 
In this chapter Pigou notes the possible restraining effect of the countervailing 
power of opposed monopolies; but he doubts its effectiveness in protecting the 


and on tenable grounds held that markets might still exercise adequate 
control, while permitting desirable progress. Perfect competition must 
fail as a useful policy norm not merely because markets do not operate 
in that way but because we would not want them to. The views to be 
corrected now by theories of a "new competition" that is "workable," 
or even by theories of "countervailing power," are rather those of fol- 
lowers of Chamberlin who fell into the bad habit of equating competi- 
tion with pure competition, of confusing theoretical benchmarks with 
policy norms, of expecting highly monopolistic behavior in most mar- 
kets where competitors are few. 

Undoubtedly the study of markets has been revitalized in the last 
quarter-century. New theory has suggested what to look for, industry 
characteristics have been revealed with new significance, new insights 
into policy have been gained. But still, in the field of practical policy, 
these developments have worked little effective change; nor is it clear 
that they will. It has seemed useful, for instance, with competition and 
monopoly commingling over a wide range, to devise means of measuring 
the degree of monopoly in markets. Ingenious techniques have been 
contrived in the abstract, and there has been some attempt to apply 
them, especially in the case of concentration indexes. But one easily 
agrees with Machlup when, after reviewing these efforts, he concludes 
that such measurement is "even conceptually impossible," quite apart 
from its applicability. 38 Oligopoly theory seems less promising now than 
it once did, not only because of its profusion of elements but because 
factors other than numbers are seen to be widely significant. 39 It is al- 
most shocking to recall the view of commentators after the Tobacco 
decision (1946) that this theory had therewith been made part of judicial 
standards and might properly dominate them in such cases. 

The nature of policy problems forces us back toward the looser ap- 
proach of earlier economists, and indeed of competent lawyers and 
judges. Even if we could measure degrees of deviation from pure com- 
petition, we would accomplish little unless pure competition were the 
market condition really desired — the condition that would promote a 
balanced achievement of diverse economic goals; and surely it is not. 
And even if we had a significant measurement, related to a truly opti- 
mum market norm, the policy question would remain: In a society in 
which ideal blueprints never materialize, what degree of departure from 
the norm is reasonably acceptable, in light of political as well as eco- 
nomic factors? More theory and more research will aid us; but there can 
be no answer except through the kind of experienced judgment always 
relied on in such matters. 

38 Fritz Machlup, The Political Economy of Monopoly (Baltimore, 1952), chap. 
12, esp. pp. 526-28. 

39 See, for instance, Carl Kaysen in the National Bureau of Economic Research 
volume, Business Concentration and Price Policy (Princeton, 1955), p. 118. 


Views differ greatly as to the desirable form and rigor of antitrust 
policy; but the differences do not really spring from a theoretical cleav- 
age. They arise, as in the past, from dissimilar appraisals of incommen- 
surable goals and market factors as seen in a framework that remains 
about the same. Economists who stress the nice equating of marginal 
results are more alarmed by monopoly elements than are economists who 
stress productivity and progress. The former have also a stricter idea of 
what reasonable profit means. Such groups may view differently the 
contribution of great firms in lowering costs and improving products, 
the competitive potency of product and technical substitutes, the need 
of market restraints to induce innovation and to prevent harmful price 
cutting where reserve capacity accompanies growth. But these grounds 
for disagreement are as explicable in the theory of Clark and Marshall 
as of today. 

To this point our theme has been developed without mention of John 
Maurice Clark, son of John Bates and leading formulator of the reasons 
why competition may be effective in an economy in which monopoly 
elements are common. In his well-known paper, "Toward a Concept of 
Workable Competition," Clark, it seems, was not trying to close a gap 
caused by failure of the older theory, but was concerned rather with 
recent refinements of the competitive model which, he said, "may serve 
as a starting point of analysis" but which, when used as a guide in ap- 
proaching policy have "seemed at times to lead to undesirable results. 
. . ." 40 In a sense he bridged the periods by paralleling the exacting mod- 
ern idea of pure competition with an equally sophisticated conception 
of the realizable and acceptable working of markets, and thus formu- 
lated with added fullness and precision a basis of policy toward which 
his father and his father's contemporaries were moving. 

Elsewhere J. M. Clark has said, in writing of his father: "What may 
reasonably be asked of the theorists of the current generation is that 
they integrate their findings with those elements of the thought of the 
preceding generation which have enduring value, and which they tend 
to neglect." 41 The present theory of pure competition and of departures 
from it grows naturally out of the older static analysis of markets; pres- 
ent theories of workable competition, even when stretched to make room 
for elements of countervailing power, likewise particularize older think- 
ing regarding feasible market operation under dynamic conditions. 
Analysis of this side of modern capitalism requires no revolution in eco- 
nomic thought. 

40 American Econo?nic Review, Vol. XXX (Proceedings of the American Eco- 
nomic Association, 1940), p. 241; reprinted in Readings in the Social Control of 
Industry (Philadelphia, 1942), p. 453. 

41 "John Bates Clark," in H. W. Spiegel, ed., The Development of Economic 
Thought (New York, 1952), p. 612. 



Public Policy and Business Size* 


The belief that bigness in business raises a problem of public policy- 
rests upon the postulate that the enterprise which is big is power- 
ful. A direct relation between bigness and power is taken as a starting 
point for the following discussion. 

The idea that big enterprises are powerful has led naturally, if not 
inevitably, to the belief that either their power should be limited or else 
its exercise should be controlled. In a democracy, since citizens are re- 
garded as equal, there is suspicion of those who rise to places of political 
power. The powerful are made responsible to the group or are sub- 
jected to a system of checks and balances designed to make sure that 
their authority is narrowly circumscribed. So long as these views prevail 
in politics, they will be expressed also in political policies toward eco- 
nomic affairs. A democratic state will necessarily adopt policies designed 
to limit the concentration of the control over business or to curb the 
business policies which may grow out of such concentration. 

Throughout economic and political life the existence of power results 
in alarm about the effects of the power upon those who are not power- 
ful. Concern about the power of large enterprises has been expressed on 
three different levels. First, and simplest, it is asserted that power may 
result in abuses of power, such as charging extortionate prices or un- 
fairly bludgeoning competitors. Such abuses are conceived as an out- 
growth of the policies of large enterprises and therefore as amenable to 
public action designed to provide incentives for a change in business 
behavior. Second, it is asserted that the power of large enterprises brings 
about an unfortunate distortion in the performance of the economic 
system because it leads to such characteristics as rigidity of prices and 
relaxation of drives toward maximum efficiency. Effects of this kind are 
conceived as the unintended by-products of the immediate policies of 
powerful enterprises and hence as lying largely beyond conscious con- 
trol by such enterprises. Third, it is asserted that an economic system in 
which power has been unduly concentrated suffers from an impairment 

* The Journal of Business of the University of Chicago, Vol. XXIV (1951), 
pp. 280-92. (Copyright, 1951, University of Chicago.) Reprinted by courtesy of the 
University of Chicago Press and the author. 

t University of Chicago. 



of some of its most important institutions, such as the opportunity for 
newcomers to undertake new ventures and the existence of a large num- 
ber of points at which decisions are made by persons who are relatively 
independent and very nearly equal in status. Such effects are conceived 
as the aggregate result of the power of various business enterprises and 
therefore as beyond control by any single person or enterprise that con- 
tributes to them. 

Persons who are concerned primarily about abuses of power, that is, 
about the first of these three effects, are likely to regard problems of 
bigness as matters depending primarily upon codes of business conduct. 
They are likely to think, with Theodore Roosevelt, that there may be 
"good trusts" and "bad trusts." Persons who are primarily interested in 
the effects of bigness upon the functioning of the economic system and 
upon the institutional arrangements therein are likely to believe that the 
policy problems created by bigness are largely independent of the point 
of view of those who manage large enterprises and cannot be substan- 
tially altered by such changes of business conduct as lie within the 
discretion of managements. Hence they are likely to think that bigness 
in business may go too far even if large enterprises are managed with 
ability and public spirit. 

Such is the setting of criticism in which public policy toward large 
enterprises must be worked out. There are also certain preconceptions 
about the scope of public policy in this field which, in my opinion, are 
shared by most of us. Among them, I think, are these six. 

1. The problems arising out of bigness are not primarily those of the 
personal character of managers of large enterprises. Even the abuses of 
power which bigness may produce are not usually due to bad personal 
character. Of course, certain business abuses may spring from lack of 
honesty, public spirit, or intelligence in a strategically placed manager. 
For the most part, however, business managements compare favorably 
with the rest of the community in these qualities. Moreover, the conduct 
of a large enterprise is a complex affair, affected by the decisions of 
many individuals, guided by policies developed in groups rather than by 
individuals, and determined in large part by the nature of the concern 
and its environment rather than by personal preferences. 

2. Some degree and kind of bigness is with us to stay. A complete 
atomization of the business structure is made impossible by the tech- 
nology of mass production and by the fact that certain forms of speciali- 
zation are possible only if a given degree of size is attained. 

3. Not all bigness is bad. Some degrees and kinds of bigness are nec- 
essary, or at least advantageous, in giving effect to the possibilities of 
modern industrial technology and in thus reducing costs and enhancing 

4. Not all bigness is good. A concern may grow bigger than is tech- 
nologically necessary and may attain such a size as to possess monopoly 


power and use that power to the detriment of its suppliers, its customers, 
and the community at large.